millennium cohort study - cls · this was to facilitate robust study of the effects of disadvantage...
TRANSCRIPT
Millennium Cohort Study
Sixth Survey 2015-2016
User Guide (First Edition)
Edited by Emla Fitzsimons
(With contributions from Vilma Agalioti-Sgompou, Lisa
Calderwood, Emily Gilbert, Lucy Haselden, Jon Johnson, Kate
Smith and the Millennium Cohort Team)
February 2017
1
Contents
Contents…………. ..................................................................................................................... 1
1. Background…………………………………………………………………………………………5
1.1 The Millennium Cohort Study........................................................................................... 5
1.2 Overview of MCS6 ........................................................................................................... 6
1.3 The sample… ................................................................................................................... 6
1.3.1 Birth dates ................................................................................................................. 6
1.3.2 Stratification….. ......................................................................................................... 6
1.3.3 Clustering……. .......................................................................................................... 7
1.3.4 Drawing the sample… ............................................................................................... 7
1.3.5 The original sample size… ........................................................................................ 8
1.3.6 MCS sample Sweeps 2 to 5 ...................................................................................... 8
2. The MCS 6 sample and response ………………………………………………………………8
2.1 MCS6 sample ............................................................................................................... 8
2.2 MCS6 response…. ....................................................................................................... 9
3. Survey development and contents.…………………………………………………………….9
3.1 Development and piloting of MCS6 ............................................................................. 9
3.2 Content……………………………………………………………………………………..10
4. Fieldwork.……………………………………………………………………………………….. ..11
4.1 Briefings… .................................................................................................................. 11
4.2 Fieldwork timetable………………………………………………………………………. .11
4.3 Languages………………………………………………………………………………....12
5. Data conventions ................................................................................................................. 14
5.1 Variable names ............................................................................................................. 14
5.2 Variable labels ................................................................................................................ 14
6. The household grid and the household questionnaire ........................................................ 16
6.1 Background and introduction ......................................................................................... 16
6.1.1 What is the household grid? ................................................................................... 16
6.1.2 How is the household grid information collected? .................................................. 16
6.2 Contents of the household grid and household questionnaire ...................................... 16
6.2.1 What information is collected in the household grid? ............................................. 16
6.2.2 What other information was collected in the household questionnaire and where is
it stored?.........................................................................................................................17
6.3 Data format .. .................................................................................................................. 17
2
6.3.1 The household grid in previous sweeps ................................................................. 17
6.3.2 MCS1 and MCS2 .................................................................................................... 17
6.3.3 Merging household grids ......................................................................................... 17
6.3.4 Known issues and data cleaning ............................................................................ 17
6.3.5 Derived variables ..................................................................................................... 17
7. Overview of parent questionnaires ...................................................................................... 19
7.1 Background and introduction ......................................................................................... 19
7.1.1 What are the ‘main’ and ‘partner’ interviews? ......................................................... 19
7.1.2 How were the ‘main’ and ‘partner’ identified at MCS6? ......................................... 19
7.1.3 Are the ‘main’ and ‘partner’ the same people in all sweeps? ................................. 19
7.2 Baseline numbers .......................................................................................................... 20
7.3 Contents of the main, partner and proxy partner interviews ......................................... 20
7.3.1 Topics covered in main and partner questionnaires ............................................... 20
7.3.1.1 Income ............................................................................................................. 22
7.3.1.2 Banded data .................................................................................................... 23
7.3.1.3 Missing income data (item non-response) ...................................................... 23
7.3.1.4 Imputation of missing and continuous income from banded data .................. 23
7.3.1.5 Income derived variables................................................................................. 24
7.3.1.6 Equivalisation ................................................................................................... 25
7.3.1.7 Construction of the poverty indicator ............................................................... 25
7.4 Scales…….. ................................................................................................................... 25
7.4.1 Big Five personality traits ........................................................................................ 25
7.4.2 Strengths and Difficulties Questionnaire (SDQ) ..................................................... 26
7.4.3 Kessler 6 scale ........................................................................................................ 28
7.4.4 Alcohol questions (AUDIT-PC) ............................................................................... 29
7.5 Unfolding brackets ......................................................................................................... 29
7.6 Feed forward data .......................................................................................................... 30
7.7 Data structures and data handling issues ..................................................................... 30
7.7.1 Parent interview file and parent derived file ............................................................ 30
7.7.2 Parent CM file .......................................................................................................... 30
7.7.3 Proxy partner interview file ...................................................................................... 30
7.8 Known issues and data cleaning ................................................................................... 30
7.8.1 Edits made by interviewers or the fieldwork agency .............................................. 30
7.8.2 Derived variables .................................................................................................... 31
8. Overview of cohort member questionnaire ......................................................................... 33
8.1 Background and introduction ......................................................................................... 33
8.1.1 The cohort members ............................................................................................... 33
3
8.1.2 Twins and triplets .................................................................................................... 33
8.2 Baseline numbers .......................................................................................................... 33
8.3 Contents of the cohort member interview ...................................................................... 35
8.3.1 Main topics .............................................................................................................. 35
8.3.2 Scales.……………………………………………………………………………………36
8.4 Data structures and data handling issues ..................................................................... 36
8.5 Known issues and data cleaning ................................................................................... 36
8.5.1 Edits made by interviewers or the fieldwork agency .............................................. 36
8.5.2 Derived variables…. ................................................................................................ 36
9. Cognitive assessments…. ................................................................................................... 36
9.1 Background and introduction… ..................................................................................... 36
9.2 Cambridge Gambling Task… ..................................................................................... 36
9.3 Word activity ............................................................................................................... 37
9.4 Consent……….. ......................................................................................................... 37
9.5 Baseline numbers…. .................................................................................................. 37
9.6 Data conventions…. ................................................................................................... 37
9.7 Derived variables…. ................................................................................................... 37
10. Physical measurements… ................................................................................................. 38
10.1 Background to physical measurements on MCS.. ...................................................... 38
10.2 Height……… ................................................................................................................ 38
10.3 Weight and body fat ..................................................................................................... 38
10.4 Weight only… ............................................................................................................... 38
10.5 Consent…. ................................................................................................................... 39
10.6 Baseline numbers… ..................................................................................................... 39
10.7 Data format ................................................................................................................... 39
10.8 Derived variables ......................................................................................................... 39
10.9 Reference cut-offs ........................................................................................................ 39
11. Accelerometry …………………………………………………………………………………..42
11.1 Background and introduction… ................................................................................... 42
11.2 Device use… ................................................................................................................ 42
11.3 Subsampling… ............................................................................................................. 42
11.4 Days worn…….. ........................................................................................................... 42
11.5 Consent………. ............................................................................................................ 42
11.6 Baseline numbers ........................................................................................................ 43
12. Time-use diary ................................................................................................................... 44
12.1 Background and introduction ....................................................................................... 44
12.2 Days completed ........................................................................................................... 44
4
12.3 Modes…… ................................................................................................................... 44
12.4 Consent…. ................................................................................................................... 45
12.5 Baseline numbers ........................................................................................................ 45
13. Saliva………….................................................................................................................46
13.1 Background and introduction ....................................................................................... 46
13.2 Saliva collection protocols ............................................................................................ 46
13.3 Consent….. .................................................................................................................. 46
13.4 Storage….. ................................................................................................................... 46
13.5 Baseline numbers ........................................................................................................ 47
14. Response and weights at Sweep 6 ................................................................................... 48
14.1 Response in MCS ........................................................................................................ 48
14.2 Predicting response at MCS6 ...................................................................................... 51
15. Available datasets .............................................................................................................. 53
16. Accompanying documents ................................................................................................ 55
16.1 Questionnaires ............................................................................................................. 55
16.2 Technical Reports ........................................................................................................ 55
16.3 Guides ……................................................................................................................ 55
17. Appendix A: Response and weights ................................................................................. 56
Imputations….. ................................................................................................................... 56
Replacement of missing values ......................................................................................... 56
Multiple imputations ........................................................................................................... 56
No imputation required ...................................................................................................... 57
18. References…… ................................................................................................................. 58
5
1. Background
1.1 The Millennium Cohort Study The Millennium Cohort Study (MCS) is a multi-disciplinary research project following the lives
of an original 18,818 children born in the UK in 2000-01. The sample was augmented in early
childhood with a further 701 children born in the same period who had been missed previously,
taking the total sample to 19,519 (note, there is no sample refreshment by immigrants). It is
the most recent of Britain’s world-renowned national longitudinal birth cohort studies. The
study has been tracking the Millennium children through their early childhood years and plans
to follow them into adulthood. It collects information directly from the children, their resident
parents and, in two of its sweeps, older siblings. The MCS covers such diverse topics as
parenting; childcare; schooling and education; daily activities and behaviour; cognitive
development; child and parent mental and physical health; employment and education;
income and poverty; housing, neighbourhood and residential mobility; and social capital,
ethnicity and identity.
The six surveys of MCS cohort members carried out so far have built up a uniquely detailed
portrait of the children of the new century. The sixth, Age 14, survey, which is the subject of
this user guide, was a major initiative, with important new additions to the study including: the
collection of saliva samples from cohort members and biological parents, enabling studies of
genetic influences on life course events and trajectories; physical activity and sedentary
behaviour, through wrist-worn accelerometers; and detailed time-use information, through
mobile phone apps/online completion. These new additions were alongside a much extended
cohort member questionnaire, and continuing parent questionnaires.
To date there have been six surveys:
MCS1 – the first sweep took place when cohort members were around 9 months old,
between June 2001 and January 2003.
MCS2 – the second sweep took place when cohort members were 3 years of age,
between September 2003 and April 2005.
MCS3 – the third sweep took place when cohort members were 5 years old; between
February 2006 and January 2007.
MCS4 – the fourth sweep took place when cohort members were 7 years old; between
late January 2008 and February 2009.
MCS5 – the fifth sweep took place when cohort members were 11 years old, between
January 2012 and February 2013.
MCS6 – the sixth, most recent, sweep took place when cohort members were 14 years
old, between January 2015 and April 2016.
Funding of MCS6
The sixth sweep of the Millennium Cohort Study was core-funded by the Economic and Social
Research Council (ESRC), and co-funded by the following consortium of government
departments: Department for Education, Department of Health, Ministry of Justice, Home
Office, Department for Transport, Department of Work and Pensions, Welsh Government and
Department for Employment and Learning (Northern Ireland).
6
1.2 Overview of MCS6 The sixth sweep of the Millennium Cohort Study was carried out when the cohort members
were 14 years old. As 14 is a key transitional age, the sweep was purposefully ambitious in
the breadth and scope of its contents. It included:
an interview (CAPI [computer-assisted personal interview] and CASI [computer-
assisted self-interview]) with the main parent and partner (where relevant)
a self-completion interview with the cohort members
cognitive assessments for the main parent, the partner and the cohort member
DNA collection of the cohort member and natural parents in the household
physical measurements of the cohort member
placement of a time-use diary with the cohort member
placement of an accelerometer with the cohort member.
1.3 The sample The original MCS sample covered children from all four countries of the UK who were eligible
for child benefit1 and were 9 months old at the time of the first sweep. It used a stratified,
clustered random sample design and oversampled from areas that were disadvantaged or
had high ethnic minority populations. This was to facilitate robust study of the effects of
disadvantage on children, as well as analysis of different ethnic groups.
1.3.1 Birth dates The cohort members were sampled from a population born across a 16-month period. This
not only allowed for season of birth to be taken into account in analysis, but also had the
practical advantage of allowing for a longer, less intense and more manageable fieldwork
period.
In England and Wales – the sample was drawn from the population of children born
between 1 September 2000 and 31 August 2001.
In Scotland and Northern Ireland – the sample was drawn from the population of
children born between 24 November 2000 and 11 January 2002.
1.3.2 Stratification In England and Wales, the population was divided into three strata:
The ethnic minority stratum was comprised of children living in wards where the
proportion of ethnic minorities in that ward in the 1991 Census was at least 30 per
cent.
The disadvantaged stratum was comprised of children living in wards, other than
those falling into the ethnic minority stratum, which fell into the poorest 25 per cent of
wards according to the Child Poverty Index for England and Wales.
The advantaged stratum comprised children living in wards other than the two
described above.
In Wales, Scotland and Northern Ireland there were only two strata (because of the low
percentages of ethnic minority groups, at around 1 per cent of the population):
1 Child Benefit claims covered virtually all of the child population except those ineligible due to recent or temporary immigration status.
7
The disadvantaged stratum was composed of children living in wards (known as
‘Electoral Divisions’ in Wales) that fell into the poorest 25 per cent of wards according
to the Child Poverty Index.
The advantaged stratum was made up of children living in other wards in these
countries.
It is important to bear in mind that both the ethnic minority indicator and the Child Poverty
Index are area-level measures. That means the design will be useful for identifying those who
are disadvantaged or from an ethnic minority background – for those who live in areas with
others from a similar background – but will be less well placed to identify those who are likely
to be part of these groups but do not live in areas with similar people. Indeed, focusing on
families in poverty, Plewis (2007) found that in England in 1998, about 37 per cent of
disadvantaged families with children under 16 were living in advantaged wards; 54 per cent
were in disadvantaged wards; and 10 per cent were in ethnic minority wards.2
1.3.3 Clustering
The sample was clustered by characteristics of electoral wards. Clustering is efficient, and it
is more cost-effective to draw a cluster sample of specific areas than to sample the whole UK.
It also helps in keeping fieldwork costs down as it enables interviewer workloads to be
concentrated, thereby reducing travel costs. Moreover, from an analysis perspective,
clustering brings the neighbourhood context into the picture, as having multiple respondents
in the same areas allows researchers to better understand area effects. Another advantage of
the cluster design is that data from the census and other sources can be matched at the
electoral ward level. However, a drawback of cluster sampling is that estimates are less
precise than those obtained from a simple random sample.
1.3.4 Drawing the sample The sample was randomly selected within each of three strata in each country, producing a
disproportionately stratified cluster sample. This means that the sample is not self-weighting,
and so weighted estimates of means, variance etc. are required (Plewis 2007).
Once the sample wards were selected, a list of all children turning 9 months old during the 16-
month survey window and living in those wards was generated from the Child Benefit (CB)
register provided by the then Department for Social Security (DSS), now the Department for
Work and Pensions (DWP). At that time, CB was a universal provision, payable (usually to the
mother) from birth. The DWP wrote to all eligible families asking the CB recipient to opt out if
they did not want to be included in the survey. An opt-out procedure tends to be more inclusive
of marginal and low literacy respondents than an opt-in procedure, and also results in higher
response rates. The DWP withdrew sensitive cases from the issued sample. These included
families where children had died or had been taken into local authority care by that point, or
where there was an investigation into benefit fraud within the family. In addition, if families had
already taken part in the DWP’s Families and Children Survey (FACS), they were excluded
from the sample.3
Because the CB records did not include all families who had moved into the sample wards as
the child approached 9 months, an additional sample was drawn using health visitors to find
eligible families who had moved into the selected areas and who had eligible children. Fifty-
six families were found in this way.
2 Percentages do not add to 100% due to rounding. 3 This affected only 40 cases.
8
1.3.5 The original sample size The MCS1 survey reached 18,552 families which, after allowing for 256 sets of twins and 10
sets of triplets, amounted to 18,818 cohort children. Six families have two singletons in the
sample. The table below shows how these respondents are distributed across the four
countries of the UK. Further details by stratum appear in the Technical Report on Sampling
(4th edition) (Plewis 2007).
Number of sample ‘wards’*
Target sample as boosted
Achieved responses**
Children Families
England 200 13,146 11,695 11,533
Wales 73 3,000 2,798 2,760
Scotland 62 2,500 2,370 2,336
N. Ireland 63 2,000 1,955 1,923
Total UK 398 20,646 18,818 18,552
*Counting amalgamations in ‘superwards’ as a single unit **All productive contacts
1.3.6 MCS sample Sweeps 2 to 5
MCS2 – The sample issued for MCS2 consisted of productive families at MCS1 and
new families that, although eligible, had not participated in MCS1. The total issued
sample was 19,870; 18,481 were productive families at MCS1 and 1,389 were new
families.
MCS3 – The sample issued for MCS3 comprised all those who had responded to the
survey at least once, i.e., to MCS1 (18,522) or to MCS2 (including 692 additional cases
who had responded to MCS2 as new families). There were 19,244 families potentially
eligible for inclusion in the issued sample; however, 718 families were not issued to
the field due to ineligibility (death or emigration), permanent refusal or sensitive family
situations.
MCS4 – The sample for MCS4 was the same as for MCS3 (i.e., those who had
responded at least once to MCS1 and MCS2). There were 19,244 families potentially
eligible for the survey. However 2,213 cases were not issued to the field due to
ineligibility from death or emigration, permanent refusal or sensitive family situations.
MCS5 – The sample for MCS5 was the same as for MCS3 and MCS4 (i.e., those who
had responded at least once to MCS1 and MCS2). There were 19,244 families
potentially eligible for the survey. However, 2,581 were not issued to the field due to
ineligibility from death or emigration, permanent refusal or sensitive family situations.
Full details on the samples and responses for each of these sweeps can be found in their
respective user guides.
2. The MCS 6 sample and response
2.1 MCS6 sample Potentially eligible families – There were 19,243 families potentially eligible for MCS6 (one
less than in previous waves as one family was identified as having had duplicate records in
previous waves).
Not issued families – 3,828 families were not issued into the field (due to death or emigration;
permanent refusal; untraceability; sensitive situations).
Issued cases – 15,415 cases were issued into the field.
9
2.2 MCS6 response
MCS6 overall response is shown in the table below:
Outcome code Number of families Percent
Productive 11,726* 60.9
Refusal 6,109 31.7
Other unproductive 115 0.6
Ineligible 668 3.5
Untraced 550 2.9
No contact 75 0.4
*Note that 12 productive families had lost information on the household grid. As a
consequence they have been removed from the deposited data as we have no information on
which parental figures are eligible or which responded. Thus the number of families available
for analysis is 11,714.
3. Survey development and contents
3.1 Development and piloting of MCS6 Development work on the MCS was extensive and covered the elaboration of survey contents,
instruments and materials as well as study engagement and branding. It also included a pilot
and a dress rehearsal which tested all aspects of the survey. Details of all the development
phases are provided below:
Participation and engagement – a number of qualitative studies examined and
tested participant engagement approaches, the dynamics of family decisions about
participation, experiences of taking part and preferences for mode of communication.
Rebranding – the study underwent a major rebranding, following extensive focus
group testing, in order to make the study materials (website, mailings) more relevant
and appealing to 14-year-olds.
Qualitative pre-testing – Qualitative research was carried out with young people in
their third year of secondary school (aged 13-14) and their parents. Eight single-sex
focus groups took place in schools across England and Scotland. A range of schools
was included according to the proportion of pupils receiving free school meals (FSM),
attainment levels and whether they were located in urban or rural areas. Twelve in-
depth interviews were conducted at home with a parent and their child from a range
of ethnic backgrounds and socio-economic groups. More information about this can
be found in the MCS6 technical report and related published material.
Cognitive testing – Selected sections of the young person questionnaire were
cognitively tested in October and November 2013. Specific objectives were to test
question wording to ensure comprehension by 14-year-olds; to explore how young
people understood and interpreted the meaning of specific terms and words in the
questions; to understand the cognitive processes young people went through to
provide answers (for example, how they retrieved, derived and reported their
answers); and to provide recommendations to change the wordings of questions to
improve reliability.
Development of the time-use record – As this was a new component at MCS6, the
development of the time-use record instruments involved extensive development
work. It was led by CLS in collaboration with Ipsos Mori (IM) and the Centre for Time
Use Research (CTUR) at the University of Oxford. CLS oversaw and contributed to all
aspects of the development. IM produced the time-use record instruments and leaflets
10
and carried out the different testing phases. CTUR made a major contribution to the
instrument development, regularly advising on key research design and
implementation decisions. This covered both cognitive testing of the activity codes and
usability testing of the survey instruments (see MCS6 Technical Report). More details
on the survey instruments can be found in the time-use diary section.
Development of survey materials – In July 2014, IM conducted interviews with
young people to test the Young Person Engagement Materials developed for the dress
rehearsal. The objectives of the materials testing were to explore young people’s
understanding of the language used, particularly in more complex sections (such as
data linkage [which was subsequently dropped for the mainstage] and saliva);
examine their understanding of the images and associated connotations; and gauge
overall reactions to the materials (e.g., whether they liked them, length, etc.). Full
details of the survey materials testing can be found in the technical report.
First pilot – The first pilot survey took place between 7 February and 2 March 2014
in five locations in England, Scotland and Wales using a quota sample to ensure that
a representative cross-section of families was included. An external agency recruited
families with a child in Year 9 in England and Wales and Secondary 3 in Scotland,
aged 13 to 14. Fifty families were interviewed, ten in each area. The pilot aimed to test
the approaches to MCS6. This included testing the length of the questionnaire, the
ethical considerations and consent; testing the implementation of each study element;
and assessing the training approach and the materials used as well as the office
procedures. Further details of the first pilot can be found in the technical report.
Dress rehearsal – The dress rehearsal fieldwork took place between 4 July and 20
August 2014 in 13 locations across England, Scotland, Wales and Northern Ireland.
The sample comprised a longitudinal sample previously recruited by CLS and used
for the dress rehearsal piloting of previous waves of the study as well as a top-up
sample sourced from the National Pupil Database (NPD) in England and via schools
in Wales, Scotland and Northern Ireland. The sample was located in 13 areas. In total,
200 addresses were issued. Of these, 152 were longitudinal samples and 48 were
new families. Because the dress rehearsal was designed to fully assess every aspect
of the survey design and implementation, it mimicked the main stage procedures and
content as closely as possible.
3.2 Content The survey contained the following key elements:
1. The household questionnaire
2. The main parent questionnaire
3. The partner questionnaire (or proxy partner questionnaire), where present
4. The cohort member (referred to interchangeably as ‘young person’) self-completion
questionnaire
5. Cognitive assessments:
a. A word activity for the main, partner (if present) and cohort member
b. Cambridge Neuropsychological Test Automated Battery (CANTAB) Cambridge
cognition gambling task
6. Physical measurements (height, weight and body fat) of the cohort member
7. Saliva samples from the cohort member and natural parents (if present)
8. Time-use diary – Placed at the time of interview, to be completed by cohort member for
two specified days (one weekday and one weekend) after the interview
11
9. Accelerometry – Placed at the time of interview, to be worn by cohort member on two
specified days (one weekday and one weekend) after the interview (to match with time-
use record days).
The diagram below provides an overview of the survey elements. It also indicates average
timings for each element, mode of administration, which consents were required (and when),
and whether the element was completed during or outside of the household visit. This chart
was used in the interviewer briefings to help interviewers to understand how each of the
different household elements fitted together and to ensure that the visit was conducted as
efficiently as possible.
The contents of each of the elements can be found in the respective sections of the user guide.
4. Fieldwork Following a competitive tender process, IM was appointed to carry out the fieldwork for MCS6.
The first wave of the mainstage fieldwork began in all countries in January 2015.
4.1 Briefings All interviewers attended a three-day briefing before working on the survey. The briefings were
run by researchers from IM and CLS, members of the IM internal field team and region
managers or region co-ordinators from IM’s field force. In total, 291 interviewers completed all
three days of the briefing. The size of the briefings varied between regions and attendance
ranged between 7 and 45 interviewers.
4.2 Fieldwork timetable Fieldwork was conducted between 15 January 2015 and 30 March 2016. The fieldwork
timetable for MCS6 was driven by the requirement to interview the family during Year 9 (Year
12
S3 in Scotland and Year 10 in Northern Ireland). As at previous sweeps, the fieldwork was
compressed into school years. In England and Wales, the cohort birth dates span a single
school year. However, in Scotland and Northern Ireland the birthdates are spread over more
than one school year. In England, Wales and Northern Ireland, school year is normally
determined by date of birth. In Scotland, school year is determined by parental choice in
addition to date of birth. It is worth noting that because of fieldwork overrunning for this sweep
of MCS, 14 per cent of families were interviewed in a different school year.
Fieldwork was split into three phases and nine waves, as shown in the following table:
Wave Fieldwork dates Countries Date of birth
Date due to start
Year 9 (England
& Wales) / Year
S3 (Scotland) /
Year 10
(Northern
Ireland)
Phase 1
1 Jan 2015 – Dec 2015 England, Wales
1 Sep 2000 – 31 Aug 2001 in England and
Wales 24 Nov 2000 – 28
Feb 2001 in Scotland 1 Sep 2000 – 31 Aug 2001 in England &
Wales 24 Nov 2000 – 28
Feb 2001 in Scotland
Sep 2014 in England and Wales
Aug 2014 in Scotland
2 Feb 2015 – Feb 2016
England, Wales, Scotland, Northern
Ireland
3 Mar 2015 – Dec 2015
England, Wales, Scotland, Northern
Ireland
Phase 2 4 Apr 2015 – Feb 2016
England, Wales, Scotland, Northern
Ireland
5 May 2015 – Dec 2015 England, Wales
Phase 3
6 Aug 2015 – Mar 2016 Scotland
24 Nov 2000 – 11 Jan 2002 in Scotland 2 Jul 2001 – 11 Jan 2002 in Northern
Ireland
August 2015 in Scotland
September 2015 in Northern Ireland
7 Sep 2015 – Mar 2015 Scotland, Northern
Ireland
8 Oct 2015 – Mar 2016 Scotland, Northern
Ireland
9 Nov 2015 – Mar 2016 Scotland
4.3 Languages A breakdown of the interviews by ‘language interviewed in’ is provided in the technical report.4
Respondents in Wales were provided with all main communication materials in both languages
and were also able to choose which language they participated in. At the appointment-making
stage, families were asked if they would like any of the parent or young person elements to
be administered in English or Welsh. If the family requested the interview to be conducted in
Welsh, the address was reallocated to a Welsh-speaking interviewer.
To support participation of parents with limited English, other language materials were
provided. (They were not provided or required for young people because all cohort members
were born in the UK and therefore have good spoken English). Parents’ materials were
provided in the seven languages most commonly required at previous sweeps of the study:
4 The MCS6 Technical Report on fieldwork is available on the CLS website: http://www.cls.ioe.ac.uk/page.aspx?&sitesectionid=1109&sitesectiontitle=MCS6+(2015)
13
Arabic, Bengali, Gujarati, Hindi, Punjabi (Gurmukhi script), Punjabi (Urdu script) and Urdu.
Occasionally the main and partner respondents were unable to speak English or were
uncomfortable with completing the interview in that language. In such cases, interviewers were
instructed to find a ‘household interpreter’ or other informal interpreter to translate some of the
elements (the consent process and parent questionnaire only). If a household interpreter was
not available, the address was reallocated to a bilingual interviewer to conduct the interview.
The nature of any language support given to respondents was recorded in the main parent
questionnaire CAPI section – specifically, whether either of the parent interviews were
translated and if so, which language and who translated (including any interviews in Welsh);
and whether any translated materials were used by the main and partner respondents and if
so, which language. Further details about translations can be found in the technical report.
14
5. Data conventions
5.1 Variable names Each question name in the instrumentation is made up of four letters. Each variable name in
the data is eight characters long – made up of the four-letter question name (e.g., ETHE), two
single-letter prefixes and two single-character suffixes as follows:
[prefix1][prefix2][question name][suffix1][suffix2] where:
Prefix1: indicates the sweep; a = MCS1; b = MCS2; c = MCS3; d = MCS4; e = MCS5; f =
MCS6
Prefix2: Identifies the instrument/respondent main and partner respondents with the
prefixes m and p respectively and proxy partner interviews with x. The full list of potential
variants for prefix 2 is:
Prefix2 Instrument/respondent
P Parental respondent
X Proxy interview
H Household module completed by main or partner respondent
D Derived
C Cohort member ‘level’ data
Question name: the four-letter question name in the questionnaire
Suffix1: identifies the iteration, i.e., where the same question is repeated for different
events/individuals:
Suffix1 Iteration
0 no iteration
1 1st iteration
2 2nd iteration
3 3rd iteration
….. and so on
Suffix 2: identifies a multi-coded variable, ie, where a single question produces more
than one answer:
Suffix2 Iteration
0 no multi-code answer
A 1st iteration
B 2nd iteration
C 3rd iteration
….. and so on
Taking this together, the variable names on the dataset have the following form:
[Sweep][Instrument][Question name][iteration][multi-coding]
5.2 Variable labels Variables are labelled in a consistent manner to aid navigation within the datasets. Labels
have abbreviated descriptions to indicate sweep, instrument and position in loops, as follows:
15
Abbreviation Description
S6 Sweep 6 (NB similar abbreviations are used for Sweeps 1-5)
DV derived variable
COG cognitive assessments, e.g., word activity and Cambridge Gambling Task
PHYS physical measurements, e.g., height and weight
MC These appear at the end of labels and indicate a multi-coded question
R These appear at the end of labels and indicate an event loop
IWR These indicate the capture of an interviewer response
IWInf These indicate the capture of interviewer information
16
6. The household grid and the household questionnaire
6.1 Background and introduction
6.1.1 What is the household grid? The household grid is part of an initial household module which is administered before any
other module in the interview. It contains information about every person in the household and
includes two types of information: individual identifiers and identifying characteristics (number,
sex and date of birth), and cross-sectional variables (e.g., relationships between household
members).
The household grid contains one record for each person who has ever appeared in the
household, for each family who participated in that sweep. Each household has a unique
number (MSCID).
There is a variable which indicates for each person whether or not they were present at any
particular sweep: AHCPRS00, BHCPRS00, CHCPRS00, DHCPRS00, EHCPRS00,
FHCPRS00 for cohort members in MCS1, MCS2, MCS3, MCS4, MCS5 and MCS6
respectively, and AHPRES00, BHPRES00, CHPRES00, DHPRES00, EHPRES00 and
FHPRES00 respectively for other people in the household. These can be used to identify
people moving into, out of or back into the household by merging the household grid files from
each sweep. Details about the household grid for previous sweeps can also be found in the
respective user guides.
6.1.2 How is the household grid information collected? At MCS6 the household grid was collected as part of the household module. It can be
completed by any adult in the household (although in practice it is usually the main or the
partner). It was collected at the start of the household visit, as its contents determined who
was eligible for other elements (especially the main and the partner elements). The household
grid used data fed forward from previous waves. In this way, it was possible to check whether
each person identified as being present at Sweep 6 had been present at any of the previous
sweeps: the person completing the grid was asked to list all of the people currently present in
the household and, for each person, was asked if that person was someone whom we had
listed as living in the household previously so that they could be assigned the same person
number. If they had never been listed as living in the household before, they were assigned a
new person number.
6.2 Contents of the household grid and household questionnaire
6.2.1 What information is collected in the household grid? The household grid collected (or confirmed) the following information for each person in the
household:
who is living in the household currently, preserving the person number of those who have
appeared previously
what happened to people who were household members at the last sweep interviewed but
who are not currently present in the house
name, sex and date of birth of new people in the household (and confirmation of these
details for previously listed people)
whether each household member is a full-time or a part-time member of the household
the working status of adults (aged 16 and over)
17
relationship of each household member to the cohort member and to each other
who has the main responsibility for caring for the cohort member.
This data is stored in the file mcs6_hhgrid.
6.2.2 What other information was collected in the household questionnaire and where
is it stored? The household questionnaire covers a number of other topics. As this information covers the
household, it appears in the parent interview file:
whether the cohort member is in a care home
the country in which the interview is taking place
whether the address is the same as at the last interview and dates of any moves
repetition of some household grid variable for the main and partner respondent for
ease of use
selection of main and partner
establishment of legal parental responsibility (for consents)
confirmation of key identification and contact details given in the live sample (for cohort
members as well as main and partner, where applicable)
consent information.
6.3 Data format The data is available as one row per person (including cohort members) ever in the household
for productive families.
6.3.1 The household grid in previous sweeps
6.3.2 MCS1 and MCS2 At MCS2, the household grid was collected independently from MCS1; i.e., the MCS1 grid
was not ‘fed forward’. In subsequent sweeps the household grid was fed forward and soft
checks were applied for basic identification such as date of birth, name and sex.
6.3.3 Merging household grids The household grids can be combined across sweeps, using the family identifier (MCSID) and
the person number of non-cohort members using [X]PNUM00 (where X is A,B,C,D,E or F).
The cohort members can also be added using [X]CNUM00 (where X is A,B,C,D,E or F).
6.3.4 Known issues and data cleaning While every attempt has been made to ensure consistency across the sweeps, reporting of
relationships in particular is sometimes problematic. For instance, the relationship between
parental figures themselves will change due to changes in cohabitation, marriage and divorce
and this will have an effect on the relationship to their children and the relationship between
siblings. This means that children will shift between ‘step’ and ‘adopted’ for instance, so care
should be taken with the use of adoption as a category. Also, there are a few families where
there are more than two natural parents. We have concentrated in the data cleaning on trying
to ensure consistency between parental figures and their relationship to cohort members.
6.3.5 Derived variables Household composition variables have been produced that are comparable to previous
sweeps. These are available in the mcs6_family_derived file.
18
Variable name Description
FDRSP000 S6 DV Parent interview response summary
FDMINT00 S6 DV Main interview outcome
FDPINT00 S6 DV Partner interview outcome
FDHTYP00 S6 DV Parents/carers in household
FDHTYS00 S6 DV Summary of parents/carers in household
FDRELP00 S6 DV Relationship between parents/carers in household
FDNATM00 S6 DV Natural mother status
FDMINH00 S6 DV Natural mother in household
FDNATF00 S6 DV Natural father status
FDFINH00 S6 DV Natural father in household
FDOTHS00 S6 DV Number of siblings of cohort member in household
FDNOCM00 S6 DV Number of cohort members in household
FDTOTS00 S6 DV Number of siblings in household plus cohort members
FDNSIB00 S6 DV Natural siblings of cohort member in household
FDHSIB00 S6 DV Half-siblings of cohort member in household
FDSSIB00 S6 DV Step-siblings of cohort member in household
FDASIB00 S6 DV Adopted siblings of cohort member in household
FDFSIB00 S6 DV Fostered siblings of cohort member in household
FDGPAR00 S6 DV Grandparent of cohort member in household
FDOTHA00 S6 DV Other adult in household
FDNUMH00 S6 DV Number of people in household excluding cohort member
FDTOTP00 S6 DV Number of people in household including cohort member
19
7. Overview of parent questionnaires
7.1 Background and introduction
7.1.1 What are the main and partner interviews? At MCS6 (as at MCS 1, 2, 3, 4 and 5), there were three possible parent interviews which could
be completed with up to two different people per family: main, partner and proxy. The main
questionnaire was given to one parent or carer (usually the mother) of the young person. It
used a combination of CAPI (computer-assisted personal interview) and CASI (computer-
assisted self-interview) and included questions about the cohort member(s) as well as about
the parent/carer him or herself and the household circumstances. The partner questionnaire
was given to their partner (if resident in the household). The proxy partner interview was asked
of the main respondent about the partner in certain circumstances. Interviewers were
instructed to only conduct a proxy partner interview when the partner was not available for
interview for the fieldwork period. The questions asked of the partner were a subset of those
asked of the main; the questions asked for proxy partners were a subset of those asked of
partners.
7.1.2 How were the main and partner identified at MCS6? The main and partner respondents were established during the household questionnaire using
an algorithm within the CAPI questionnaire. The algorithm was based exclusively on
relationships between household members. In certain circumstances the interviewer could
override this selection and replace the person selected for the main interview with the one
selected for the partner interview. This might happen if, for instance, the person selected in
CAPI as the partner was the main carer, or if the mother was unwilling to take part but her
partner was, and was willing to be interviewed as the main respondent.
In most cases, the CAPI selected the mother figure to complete the main questionnaire.
However, there were notable exceptions: If the father was the only natural parent in the
household, he was chosen; if there were no parents (including step, foster and adoptive) living
with the young person, the CAPI selected the main carer and his or her partner for interview.
If the person selected as the main had a partner living in the household, that person would be
eligible to complete the partner interview.
For details of identification of main and partner in previous sweeps see Hansen (2008).
7.1.3 Are the main and partner the same people in all sweeps? Not necessarily, although the criteria for the selection of main and partner were the same (or
very similar) in each sweep, with main generally being the natural mother where possible.
However, in some cases the household composition changed over time; in others, the CAPI
selection was overridden by the interviewers. The table below shows the proportion of
households with the same main respondent, as at their previous interview.
Sweep Same main respondent
MCS2 98.7%
MCS3 96.7%
MCS4 96.5%
MCS5 96.1%
MCS6 95.5%
20
It is possible to check the person number of the main for MCS6 in FPNUM00. The person
number for the main has been stored in similar variables for previous sweeps. The person
number assigned to a household member is consistent throughout all prior sweeps.
7.2 Baseline numbers
MCS6 Proportion of households with main and partner interviews
Frequency Percent
1. Main respondent in person (no-one eligible for partner)
2,853 24.4
2. Main and partner respondent interviewed in person
7,230 61.7
3. Main respondent interviewed in person; partner by proxy
334 2.9
4. Main respondent interviewed in person; Partner eligible but no response
1,170 10.0
5. No main interview; partner interviewed in person
33 0.3
6. No main interview; no eligible partner 28 0.2
7. No parent interview 66 0.6
All productive households 11,714 100
Note: Percentages may not sum to 100 due to rounding.
MCS6 Sex and relationship of main respondent to cohort member
7.3 Contents of the main, partner and proxy partner interviews
7.3.1 Topics covered in main and partner questionnaires The table below shows the main topics covered in the MCS6 main, partner and proxy
interviews. All were through interview (CAPI), with the exception of those under ‘self-
completion’, which were done on their own (without interviewer) through CASI.
Frequency Percent
Female Natural mother 10,782 98 Other 225 2 Male
Natural father 664 94
Other 43 6 All main respondents 11,714 100
21
Broad topic Sub-topic Main Partner Proxy
Family context Marital status
Languages spoken at home
Ethnic group
Reasons for separation from previous partner
Relationship/contact with absent parent of cohort member
Cohort member contact with natural parents
Own parents
Own childhood circumstances
Periods when cohort member was not living with main
Education and schooling
School year
Details of the school the cohort member currently attends (or reasons for not currently attending school)
Reasons for attending a fee-paying or faith school
Language taught in
Details of other schools attended and reasons for leaving
Periods of absence from school
Details of any special needs
Parental aspirations for cohort member after leaving school
Parent–school communication
Details of any extra tuition
Travel to school
Free school meals
Parenting activities Eating together
Cohort member chores
Parental knowledge of cohort member going out
Young person’s health
Details of cohort member’s longstanding illnesses
Eyesight problems
Speech problems
Details of atopic conditions
Details of certain communicable diseases
Behavioural disorders
Accidents and injuries
Admissions to hospital
Vaccinations
Parent’s health
Parent’s general health
Details of any longstanding illnesses
Smoking
Physical health
Employment and income
Details of current jobs (including second jobs)
22
Employment and unemployment history since last interview
Hours worked
Paid and unpaid overtime
Activity if unemployed
Reasons for absence from employment
Current pay
Partner’s current pay
Benefits received
Pensions received
Other income
Assets
Debts
Financial wellbeing
Qualifications since last interview
Literacy and numeracy
Housing Details of current accommodation
Cohort member’s bedroom
Tenure
Rent
Housing benefit
Mortgage
Reasons for moving to this address
Housing history since last interview
Periods of homelessness
Car ownership
Local area
Other matters Religion
Time spent with children
Self-completion Personality traits (Big Five)
Relationship with cohort member
Cohort member’s ability to control emotions
Own mental health
Own alcohol consumption
Cohort member’s alcohol consumption
Own drug consumption
Own happiness
Violent relationships
Partner living outside the house
Life satisfaction
Contact information Address and contact details used by the MCS team to maintain contact for future waves
Strengths and Difficulties Questionnaire (SDQ)
A strengths and difficulties questionnaire on paper
7.3.1.1 Income
Income has been collected at each wave of the MCS through two banded questions
administered to two-parent and single-parent families respectively. This section
23
describes the collection of income measures in the survey, and the derivation of the
income-derived variables at MCS6 and poverty indicator.
7.3.1.2 Banded data
Respondents were shown a card with weekly, monthly and annual bands of total take-
home income from all sources and earnings after tax and other deductions. These
sources implicitly included state benefits, which had been the subject of more detailed
previous questions. Note that, unlike other state benefits, there was no attempt to
ascertain the amounts of Housing Benefit and Council Tax Benefit as separate
components. Therefore, these may well have been omitted from estimates of total net
income. Bands of different sizes were used for single- and two-parent families. In the
latter, both main respondents and partners answered the banded income question.
There are, therefore, two reported values for each household in the case of two-parent
families. In 54 per cent of the households with two parents/carers, main respondents and
partners reported a different value.
7.3.1.3 Missing income data (item non-response)
Some families did not report income: the table below shows that 1,689 of MCS families
in wave 6 did not provide banded income data, and shows why.
Missing data on the banded income questions (number of families)
Number of families
Missing income data (refusal) 368
Missing income data (don’t now) 1,179
Other missing 142
Total missing 1,689
Observed number of families 11,714 7.3.1.4 Imputation of missing and continuous income from banded data
Income was imputed for the main respondents and for partners in two-parent families,
where missing, using interval regression (Stewart 1983). This method allowed us to
impute a continuous value within a band, rather than assuming that all cases in a band
had the same midpoint income. This was achieved using Stata’s INTREG command
(StataCorp 2007; Conroy 2005). INTREG fits a model of y = [dependent variable 1,
dependent variable 2] on independent variables where the dependent variable 1 was the
log lower income band and dependent variable 2 was log upper income band. Note that
the left-hand-side bound for the lowest band is 0 and the right-hand-side bound for the
top band is the 100th income percentile in the UK. The predictors are shown in the table
below.
Predictors of income in MCS6
Variable Categories
Main respondent’s age at
interview Continuous
Housing tenure
Own
Private renting
Renting from local authority or housing
association
24
Other
Person currently in work Yes
No
Sampling point type
Advantaged
Disadvantaged
Ethnic
DV interview government
region
North East
North West
Yorkshire and the Humber
East Midlands
West Midlands
East of England
London
South East
South West
Wales
Scotland
Northern Ireland
In receipt of state benefit? No
Yes
Main respondent's ethnic
background
White
Mixed
Indian
Pakistani and Bangladeshi
Black or Black British
Other ethnic group (including Chinese and
other Asian)
DV combined education
Highest level of either
respondent
NVQ level 1
NVQ level 2
NVQ level 3
NVQ level 4
NVQ level 5
Overseas quallfication only
Other including NA
Main type of
accommodation
A house or bungalow
A flat or maisonette
A studio flat
Number of children
including cohort member
1; 2; 3; 4 or more
Derived variable number of
parents /carers in
household
Two parents/carers
One parent/carer
7.3.1.5 Income-derived variables
Variable name Description
25
FOEDE000 S6 DV OECD Equivalised weekly family Income
FOEDP000 S6 DV OECD Below 60% mean indicator
FOECDUK0 S6 DV OECD Equivalised income quintiles – UK whole
FOECDSC0 S6 DV OECD Equivalised income quintiles – by country
7.3.1.6 Equivalisation
After imputation, the values for main respondents and partners were averaged for
families with two parents. This yielded a continuous income measure for each family in
MCS6. Modified OECD scales were used for equivalisation. Each scale sets the family’s
needs relative to those of a couple with no children, whose scale is set equal to 1. In the
modified OECD scale, a family of one parent and one child under 14 has a scale of 0.87;
one parent and two children 1.07; and so on. This is shown in the table below.
OECD household equivalence scales
Equivalence scales before housing cost OECD scale used
First adult (main respondent) 0.67
Spouse 0.33
Dependent child aged between 14 and 18 years 0.33
Child aged under 14 years 0.2
7.3.1.7 Construction of the poverty indicator
A binary poverty indictor was constructed based on the OECD equivalised income. The
indicator takes the value 1 if the OECD equivalised income of the household is below 60
per cent of the DWP national HBAI weekly median income, which was equal to £284 in
2014-15 (DWP 2016). Moreover, two variables containing income quintiles were
constructed (one for the UK as a whole and one for each of the four countries of the UK:
England, Wales, Scotland and Northern Ireland). The quintiles were weighted using the
overall weights in MCS6 for the UK as a whole and for ‘by-country’ analyses.
7.4 Scales
7.4.1 Big Five personality traits The Big Five personality traits, also known as the five factor model (FFM), is a model
based on common language descriptors of personality. The five factors have been defined as
openness to experience, conscientiousness, extraversion, agreeableness and neuroticism,
often listed under the acronyms OCEAN or CANOE.
The main and partner were each asked to rate how much each of the following 15 statements
applied to them using a scale of 1 to 7, where 1 is ‘does not apply to me at all’ and 7 is ‘ applies
to me perfectly’.
Question name
Question Variable name
BIGA I see myself as someone who is sometimes rude to others
FPBIGA00
BIGB I see myself as someone who does a thorough job FPBIGB00
BIGC I see myself as someone who is talkative FPBIGC00
BIGD I see myself as someone who worries a lot FPBIGD00
BIGE I see myself as someone who is original, coming up with new ideas
FPBIGE00
BIGF I see myself as someone who has a forgiving nature FPBIGF00
26
BIGG I see myself as someone who tends to be lazy FPBIGG00
BIGH I see myself as someone who is outgoing, sociable FPBIGH00
BIGI I see myself as someone who gets nervous easily FPBIGI00
BIGJ I see myself as someone who values artistic, aesthetic experiences
FPBIGJ00
BIGK I see myself as someone who is considerate and kind to almost everyone
FPBIGK00
BIGL I see myself as someone who does things efficiently FPBIGL00
BIGM I see myself as someone who is reserved FPBIGM00
BIGN I see myself as someone who is relaxed, handles stress well
FPBIGN00
BIGO I see myself as someone who has a big imagination FPBIGO00
Note, parental personality type was assessed at MCS4 using a different set of questions.
These and other scales can be found in Johnson et al. (2015).
The Big Five derived variables
Description Sub-scale Variable name
FDOPEN Openness S6 DV OCEAN
FDCONSC Conscientiousness S6 DV OCEAN
FDAGREE Agreeableness S6 DV OCEAN
FDNEUROT Neuroticism S6 DV OCEAN
FDEXTRAV Extroversion S6 DV OCEAN
Derived variables were constructed as follows:
FDOPEN = FPBIGE00 + FPBIGJ00 + FPBIGO00 FDCONSC = FPBIGB00 + FPBIGG00* + FPBIGL00 FDAGREE = FPBIGA00* + FPBIGF00 + FPBIGK00 FDNEUROT = FPBIGD00 + FPBIGI00 + FPBIGN00* FDEXTRAV = FPBIGC00 + FPBIGH00 + FPBIGM00* The sub-scales are only computed if all the component items are completed. *Denotes the
item was reversed.
7.4.2 Strengths and Difficulties Questionnaire (SDQ) The SDQ is a behavioural screening questionnaire for 3- to 16-year-olds. It measures 25 items
on psychological attributes (Goodman 1997).
At MCS6, the P4-17 - SDQ and impact supplement for the parents of 4-17-year-olds version
was used.
The respondent is asked to comment on the following statements with response options: Not
true, Somewhat true or Certainly true.
Question name
Question Variables Equivalent on SDQ
SDPF Considerate of others’ feelings FPSDPF00 SDQ item 1
SDRO Restless, overactive, cannot stay still long
FPSDRO00 SDQ item 2
SDHS Complains of headaches/stomach-aches/sickness
FPSDHS00 SDQ item 3
SDSR Shares readily with others FPSDSR00 SDQ item 4
27
SDTT Often has temper tantrums FPSDTT00 SDQ item 5
SDSP Tends to play alone FPSDSP00 SDQ item 6
SDOR Generally obedient FPSDOR00 SDQ item 7
SDMW Often seems worried FPSDMW00 SDQ item 8
SDHU Helpful if someone is hurt, upset or ill FPSDHU00 SDQ item 9
SDFS Constantly fidgeting FPSDFS00 SDQ item 10
SDGF Has at least one good friend FPSDGF00 SDQ item 11
SDFB Fights with or bullies other children FPSDFB00 SDQ item 12
SDUD Often unhappy FPSDUD00 SDQ item 13
SDLC Generally liked by other children FPSDLC00 SDQ item 14
SDDC Easily distracted FPSDDC00 SDQ item 15
SDNC Nervous or clingy in new situations FPSDNC00 SDQ item 16
SDKY Kind to younger children FPSDKY00 SDQ item 17
SD0A Often lies or cheats FPSD0A00 SDQ item 18
SDPB Picked on or bullied by other children FPSDPB00 SDQ item 19
SDVH Often volunteers to help others FPSDVH00 SDQ item 20
SDST Can stop and think before acting FPSDST00 SDQ item 21
SDCS Steals from home, school or elsewhere
FPSDCS00 SDQ item 22
SDGB Gets on better with adults FPSDGB00 SDQ item 23
SDFE Many fears, easily scared FPSDFE00 SDQ item 24
SDTE Sees tasks through to the end, good attention span
FPSDTE00 SDQ item 25
The above 25 items are divided between 5 scales:
1. Emotional symptoms
a. Complains of headaches/stomach aches/sickness
b. Often seems worried
c. Often unhappy
d. Nervous or clingy in new situations
e. Many fears, easily scared.
2. Conduct problems
a. Often has temper tantrums
b. Generally obedient*
c. Fights with or bullies other children
d. Steals from home, school or elsewhere (In MCS2: Can be
spiteful to others)
e. Often lies or cheats (in MCS2: Often argumentative with
adults).
3. Hyperactivity/inattention
a. Restless, overactive, cannot stay still for long
b. Constantly fidgeting
c. Easily distracted
d. Can stop and think before acting*
e. Sees tasks through to the end*.
4. Peer relationship problems
a. Tends to play alone
b. Has at least one good friend*
c. Generally liked by other children*
d. Picked on or bullied by other children
28
e. Gets on better with adults.
5. Prosocial behaviour
a. Considerate of others’ feelings
b. Shares readily with others
c. Helpful if someone is hurt, upset or ill
d. Kind to younger children
e. Often volunteers to help others.
*Denotes items that are reversed when generating sub-scales on behaviour.
Each of the five scales can be used alone or together to create:
1-4 when taken together generate a total difficulties score
1 and 4 create an internalising problems score
2 and 3 create an externalising conduct score
5 alone measures prosocial behaviour
SDQ derived variables
Variable name Description
FDEMOT00 S6 DV SDQ Emotional symptoms
FDCOND00 S6 DV SDQ Conduct problems
FDHYPE00 S6 DV SDQ Hyperactivity/inattention
FDPEER00 S5 DV SDQ Peer problems
FDPROS00 S6 DV SDQ Prosocial
FDEBDTAA S6 DV SDQ Total difficulties
The SDQ was also administered in Sweeps 2, 3 4 and 5. For further details see Johnson et
al. (2015).
7.4.3 Kessler 6 scale The Kessler 6 (K6) scale is a quantifier of non-specific psychological distress. It consists of six
questions about depressive and anxiety symptoms that a person has experienced in the last
30 days. It was asked as part of the main and partner self-completion questionnaires (Kessler
et al. 2003).
For each question, respondents were offered a self-report scale of five possible answers
plus don’t know/don’t wish to answer:
1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time
6. Don’t know/Don’t wish to answer
The questions are preambled by the statement: ‘The next few questions are about how you
have felt over the last 30 days’. The six questions are:
Question name Question Variable
PHDE During the last 30 days, about how often did you feel so depressed that nothing could cheer you up?
FPPHDE00
PHHO During the last 30 days, about how often did you feel hopeless?
FPPHHO00
29
PHRF During the last 30 days, about how often did you feel restless or fidgety?
FPPHRF00
PHEE During the last 30 days, about how often did you feel that everything was an effort?
FPPHEE00
PHWO During the last 30 days, about how often
did you feel worthless?
FPPHWO00
PHNE During the last 30 days, about how often
did you feel nervous?
FPPNHE00
Kessler-derived variables
Variable name Description
FPKESS00 S6 DV Kessler K6 Scale
Derived variable FPKESS00 was constructed by summing the items: PHDE (reversed), PHHH
(reversed), PHRF (reversed), PHEE (reversed), PHHW (reversed), PHNE (reversed) where
(1=4) (2=3) (3=2) (4=1) (5=0) (6=missing).
7.4.4 Alcohol questions (AUDIT-PC) The alcohol questions in the self-completion section of the main parent questionnaire are a
shortened, adapted version of the 10-item Alcohol Use Disorders Identification Test Primary
Care (AUDIT-PC), developed by the World Health Organisation.
The five questions are:
Question name
Question Variable
ALDR How often do you have a drink that contains alcohol? FPALDR00
AUND How many standard alcoholic drinks do you have on a typical day when you are drinking?
FPAUND00
AUSD How often in the last year have you found you were not able to stop drinking once you had started?
FPAUSD00
AUAC How often in the last year have you failed to do what was expected of you because of drinking?
FPAUAC00
AUCD Has a relative, friend, doctor or health worker been concerned about your drinking or advised you to cut down?
FPAUCD00
AUDIT-PC derived variables
Variable name Description
FPAUDIT S6 DV AUDIT-PC Scale
Derived variable FPAUDIT was constructed by summing the five responses. A total of 5+
indicates increasing or higher risk drinking.5
7.5 Unfolding brackets Unfolding brackets were used on a number of questions. If respondents were unable or
unwilling to give an exact amount as an answer to a particular question (for example exact
5 https://www.alcohollearningcentre.org.uk/Topics/Latest/AUDIT-PC/
30
income), unfolding brackets were used to gain an estimate of that amount. So for instance, if
the respondent could not answer a question on income, s/he was asked whether his/her
income was above or below a rounded income amount (for example £20,000). S/he could then
be asked a series of similarly structured questions in order to narrow down the amount range.
The data is currently in preparation and more details will be available when it is finished.
7.6 Feed forward data Some information was fed forward from earlier sweeps. The main was identified by a person
number which was recorded on the Computer-Assisted Interview (CAI). If the person number
for the main was associated with someone who had been the main or partner at a previous
sweep, personal information about her or him was fed forward from the last sweep they had
participated in. In addition, information that related to the cohort member or the household as
a whole could be fed forward from the previous sweep even if the main respondent was
different.
7.7 Data structures and data-handling issues We have endeavoured to make the data available in a format that is conducive to use by as
wide a range of researchers as possible. As with MCS5, we have released data at a person
level, so that it is possible to combine both individual adult respondents and those of the cohort
members themselves. The adult respondents can be joined with other data files in MCS6 and
to prior sweeps of data collection using the person number (FPNUM00) variable. In the parent
CM file, there is additionally the cohort member number (FCNUM00) variable, which allows
joining of the cohort member across data collections. The advantage of this data structure
(over that previously available for MCS1 to MCS4) is that the data can be combined
longitudinally more straightforwardly, using the person identifier (FPNUM00) and then
selecting, for instance, only natural mothers or other characteristics of interest. In addition,
because the questions for both respondents are asked identically, the computation for any
variable is simplified.
7.7.1 Parent interview file and parent derived file The parental interviews are available in the data as one row per respondent for those
questions asked about the parental figures, e.g. their employment, income, housing,
qualifications, and those asked in the self-completion.
7.7.2 Parent CM (cohort member) file Questions which are asked of the parental figures about the cohort members, the CM’s
education, health and self-completion are available in the data as one row per CM per parental
response. This also includes the strength and difficulties questions asked on paper.
7.7.3 Proxy partner interview file For those questions asked about partners who are either away or incapacitated, the person
number (FPNUM00) variable relates to the person being reported upon. The data is available
as one row per proxy respondent.
7.8 Known issues and data cleaning
7.8.1 Edits made by interviewers or the fieldwork agency Interviewers carried out most of the data editing in the field, where inconsistencies were
highlighted through ‘soft’ and ‘hard’ checks. The former did not allow entries outside a given
range (and had to be resolved by the interviewer at the time of the interview); the latter required
31
the interviewers to check and confirm what they had entered. These enabled interviewers to
clarify and query data discrepancies directly with the respondent during the interview.
Interviewers recorded in the Final Element module instances where they believed further
amendments to the data would be needed, and, in a few instances, interviewers notified Head
Office where other amendments to the data were necessary. These reports were reviewed
and, if required, an edit was proposed. The proposed edits were then signed off by CLS. If an
edit to a variable had routing implications in the questionnaire, the following approach was
taken:
In most cases where, after applying the edit, questions had been answered that should
not have been answered, the data contained in those (routed) variables was cleared.
However, there were cases where this was not done due to complexity and the risk of
introducing further problems. All such cases were agreed with CLS.
In cases where questions would have been asked but were not, the response was left
as ‘1: not applicable’. However, in some places in the script, it was beneficial to
complete these variables. All such cases were agreed with CLS.
Some edits were made to body fat and height measurements based on comments interviewers
had made in the physical measurements module. Additionally, edits were made in cases
where the body fat value and the weight value were identical; this was deemed to be
interviewer error and both values were deleted. All such edits were agreed with CLS.
All edited cases were flagged within the final data sets using a predefined set of variables,
along with any cases where data was not edited but where there may have been issues. All
such flags are shown in Figure 9.2 of the MCS 6 technical report.6
7.8.2 Derived variables A number of variables, many of which have also been calculated for previous sweeps, are
listed below. Those questions asked of the parental respondents appear in the parental
derived file, and those relating to the cohort member, e.g., the Strengths and Difficulties battery
appear in the CM derived file.
Detailed documentation on their derivation can be found in the MCS6 Derived Variable User
Guide.
Parental derived file
Variable name Description
FDSAM00 S6 DV Respondent same as at Sweep 5
FDLST00 S6 DV Respondent status at Sweep 5
FDRES00 S6 DV Respondent identity and interview status
FDREL00 S6 DV Respondent relationship to CM
FDAGI00 S6 DV Respondent age at interview
FDGAI00 S6 DV Respondent age at interview (grouped)
FPRXF00 S6 DV Proxy partner interview flag
FDWRK00 S6 DV Whether respondent is in work or not
FDEMP00 S6 DV Employment status for Standard Occupational Classification (SOC) coding
FDACT00 S6 DV Respondent economic activity status
FD17S00 S6 DV NS-SEC full version (current job)
6http://www.cls.ioe.ac.uk/page.aspx?&sitesectionid=1109&sitesectiontitle=The+age+14+survey+of+MCS+(2015))
32
FD13S00 S6 DV NS-SEC 13 category (current job)
FD07S00 S6 DV NS-SEC 7 category (current job)
FD05S00 S6 DV NS-SEC 5 category (current job)
FDEEA00 S6 DV Respondent's ethnic group (England)
FDEWA00 S6 DV Respondent's ethnic group (Wales)
FDESA00 S6 DV Respondent's ethnic group (Scotland)
FDENA00 S6 DV Respondent's ethnic group (Northern Ireland)
FD06E00 S6 DV Respondent's ethnic group - 6 category census classification (UK)
FD11E00 S6 DV Respondent's ethnic group - 11 category census classification (UK)
FD08E00 S6 DV Respondent's ethnic group - 8 category census classification (UK)
FDKESSL S6 DV Kessler K6 scale
FDOPEN S6 DV OCEAN - Openness sub-scale
FDCONSC S6 DV OCEAN - Conscientiousness sub-scale
FDEXTRAV S6 DV OCEAN - Extraversion sub-scale
FDAGREE S6 DV OCEAN - Agreeableness sub-scale
FDNEUROT S6 DV OCEAN - Neuroticism sub-scale
FPAUDIT S6 DV AUDIT-PC scale
FDNVQ00 S6 DV Respondent NVQ highest level (all sweeps)
FDACAQ00 S6 DV NVQ equivalent of highest academic level across sweeps
FDRLG00 S6 DV respondent religion - 7 category
Cohort member derived file
Variable name Description
FEMOTION S6 DV Parent-reported CM SDQ emotional symptoms
FCONDUCT S6 DV Parent-reported CM SDQ conduct problems
FHYPER S6 DV Parent-reported CM SDQ hyperactivity/inattention
FPEER S6 DV Parent-reported CM SDQ peer problems
FPROSOC S6 DV Parent-reported CM SDQ prosocial
FEBDTOT S6 DV Parent-reported CM SDQ total difficulties
33
8. Overview of cohort member questionnaire
8.1 Background and introduction
8.1.1 The cohort members At MCS6 the cohort members were at a significant age, between childhood and adulthood.
Rather than asking their parents, the Age 14 Survey focused much more on the cohort
members themselves, as they were doing more in their lives and were able to report on their
own activities, thoughts and feelings. The cohort members were asked to complete a 40-
minute CASI (self-completion) questionnaire on the interviewer’s tablet. They were
encouraged to complete the questionnaire in private. In the very few cases where cohort
members were unable to complete the questionnaire themselves, the interviewer could read
out the questions to them. However, in these cases sensitive questions were skipped. In a
small number of cases (8), households have non-twin siblings who are also cohort members
(they were born within the time period of the original sample). These can be identified in the
longitudinal family file in the DUALBFAM variable.
8.1.2 Twins and triplets In households where there was more than one cohort member resident, each cohort member
was asked to complete a questionnaire.
8.2 Baseline numbers
Number of cohort members in each household
Frequency Percent
Singletons 11,714 98.7
Twins 150 1.3
Triplets 8 0.1
All Cohort Members 11,872 100
Non-twin siblings 8 0.1
Note: Percentages may not sum to 100 due to rounding.
34
MCS6 Sex and age at interview of cohort members
Frequency Percent
Female, age
13 1,416 23.9
14 2,232 74.8
15 78 1.3
Male, age
13 1,444 24.3
14 4,420 74.3
15 82 1.4
35
8.3 Contents of the cohort member interview
8.3.1 Main topics The age 14 cohort member questionnaire covered a wide range of topics about their lives and
comprised the following broad areas.
Things that they do
What activities they do in their free time
Watching TV and using internet, computers, games
consoles
Getting pocket money or money from paid work
Their views Attitudes to gender roles and consumerism
Attitudes to activities like fighting, shoplifting, and spray
painting
School and their future What subjects they study at school
Homework
How they feel about school
Behaviour in school and truancy
Their future education and work
About them Religion and ethnicity
Their family Relationship with parents and grandparents
Parental control and discipline
Contact with parents who don’t live with them
Their friends Type of friends, and what their friends are like
How much time they spend with friends
Relationships Support from friends and family
Romantic relationships
Sexual experiences
Things they may have tried Smoking
Drinking alcohol
Illegal drugs
Gambling
Things they may have
experienced
Being bullied and bullying, including cyber-bullying
Being a victim of crime
Things they may have done Being involved with illegal and anti-social behaviour
Contact with the police
Their health What they eat and drink
Sight and hearing problems
Dental health
Sleeping habits
Their body Dieting and trying to lose weight
Puberty
How they feel How happy they feel with their lives
Moods and feelings
More about them Attitude to trust, patience and risk
36
8.3.2 Scales
Questions Topic Source
SAFF, TRSS, NCLS Social support Three items from Social Provisions Scale – short form
HHND – CONP Sexual experience
Adapted from Adolescent Sexual Activity Index (ALSPAC)
FRUT, BRED, MILK Diet Eating Choices Index
PUHG-AGMN Puberty Pubertal Development Scale
SATI-GDSF Wellbeing/self- esteem
Shortened Rosenberg Self-esteem Scale
MDSA-MDSM Mental health Moods and Feelings Scale short form
8.4 Data structures and data handling issues The data collected in the cohort member interviews is available in the CM interview file and
the CM derived file, with one row per cohort member.
8.5 Known issues and data cleaning
8.5.1 Edits made by interviewers or the fieldwork agency Please refer to section 7.8.1 for this section.
8.5.2 Derived variables These variables are available in the CM derived file.8.5.1
Variable Description
FDCE0600 S6 DV CM ethnic group classification - 6 categories
FDCE0800 S6 DV CM ethnic group classification - 8 categories
FDCE1100 S6 DV CM ethnic group classification - 11 categories
9. Cognitive assessments
9.1 Background and introduction
9.2 Cambridge Gambling Task Young people were asked to complete the Cambridge Gambling Task during the interviewer
visit. The Cambridge Gambling Task is taken from the Cambridge Neuropsychological Test
Automated Battery (CANTAB). It measures decision-making and risk-taking behaviour.
The assessment was administered using the interviewer’s tablet, and the interviewer guided the young person through the assessment using a laminated script. The young person was presented with a row of ten boxes across the top of the screen, some of which were red and some of which were blue. The young person had to decide whether a ‘token’ was hidden in a red box or a blue box. The young person started with a number of points displayed on the screen and had to decide what proportion of their points they were willing to risk on their decision. The young person had to try to accumulate as many points as possible.
37
The same assessment was completed with the young people when they were 11.
9.3 Word activity Young people and resident parents were asked to complete the word activity during the
interviewer visit. It measures respondents’ understanding of the meaning of words.
The assessment involved presenting the respondent with a list of target words, each of which
had five other words next to them. The respondent had to select, from the five options, the
word which meant the same, or nearly the same, as the target word. Each respondent had a
list of 20 target words, and the lists of words were different for the young person, main parent
and partner. The assessment was carried out on the interviewer’s tablet.
The words used in the word activity are subsets of those used in a vocabulary assessment in
the 1970 British Cohort Study (BCS70) Age 16 survey, which took place in 1986. The words
used in the BCS70 assessment come originally from the standardised vocabulary tests
devised by the Applied Psychology Unit at the University of Edinburgh in 1976.
9.4 Consent As the young people were still children, parents were gatekeepers to their participation.
Information about the cognitive assessments was provided to both parents and young people
in advance of the interviewer visit in the form of a booklet, which contained information on why
the survey included cognitive assessments and briefly explained what each assessment was.
Parents were asked to provide written consent for their own participation. Parents were also
asked to provide written consent for the interviewer to approach their child to participate, and
young people gave verbal consent using a structured consent form, which the interviewer read
out, and then signed.
9.5 Baseline numbers
Proportion of young people who completed the Cambridge Gambling Task and word activity
No. %
Cambridge Gambling Task 10,842 91.3
Word activity 10,921 92.0
Proportion of main parents and partners who completed the word activity
No. %
Main parent 11,057 95.4
Partner 6,869 94.6
9.6 Data conventions For the word activity exercise, different variables have been output for the main (FPMCOG0x)
and partner (FPPCOG0x) respondent (as they were given different word lists, the labelling
makes this clear).
9.7 Derived variables
Variable Description
FCGTOUTCM CGT Test outcome
FCGTTTIME CGT Test duration (seconds)
FCGTDELAY CGT Delay aversion
38
FCGTDTIME CGT Deliberation time
FCGTOPBET CGT Overall proportional bet
FCGTQOFDM CGT Quality of decision-making
FCGTRISKA CGT Risk adjustment
FCGTRISKT CGT Risk-taking
FCWRDSC CM Word activity score out of 20
10. Physical measurements
10.1 Background to physical measurements on MCS At age 14 height, weight and body fat measurements were taken by the interviewer for each
cohort member. Physical measurements have been collected from cohort members since the
age of 3. Height and weight measurements have been taken at each survey (ages 3, 5, 7, 11
and 14). In addition, waist measurements were taken at ages 5 and 7, and body fat
measurements were taken at ages 7 and 11.
Interviewers were accredited to take the physical measurements at age 14 in order to ensure
accurate and consistent measurements. Reasons for not being able to take any measurement
and circumstances that applied to measurements were recorded by the interviewer in CAPI.
The data collection instrument limited height to between 120cm and 200cm, and weight to
between 20kg and 120kg (even where interviewers confirmed the value outside the range was
correct and re-entered it).
10.2 Height The height measurement was taken by the interviewer using a Leicester height measure. The
interviewer used a Frankfort Plane card to check that the cohort member’s head was in the
correct position. The measurement was taken in metres and centimetres and rounded down
to the nearest completed millimetre. Cohort members had to be able to stand unaided in order
for the height measurement to be taken.
10.3 Weight and body fat Weight and body fat measurements were taken together using Tanita™ scales. The body fat
measurement was taken by sending a weak electronic current around the body from one foot
to the other. The scales measure the amount of resistance encountered by the current as it
travels round the body. As muscle and fat have different levels of resistance, the scales use
this to calculate body fat percentage. Weight measurements were recorded in kilograms and
body fat was recorded as a percent, both to one decimal place. The scales required the cohort
member’s height and age to be entered before the measurements could be taken, so height
had to be measured first.
10.4 Weight only If the cohort member did not want their body fat measurement to be taken or if the body fat
measurement could not be taken (e.g., no height measurement was possible), the Tanita™
scales could be operated to take weight only. This was measured in kilograms to one decimal
place.
39
10.5 Consent Before taking any of the physical measurements, interviewers sought written consent from the
parent/guardian (recorded at CHIC) and, if they agreed, verbal consent was obtained for each
measurement from the young person and confirmed in CAPI (recorded at CHAC). After the
measurements were taken the interviewer asked the young person if they would like a record
of any of their measurements. If so, these were recorded on a measurement card and given
to the young person.
10.6 Baseline numbers Measurement Male Female
Height 5,699 5,701
Weight 5,644 5,506
Body fat 5,614 5,486
Missing 247 225
Total 5,946 5,926
10.7 Data format Data from the physical measurements module is available in the CM measurement file, and
the derived variables are held in the CM derived file. Both of these contain one row per cohort
member.
10.8 Derived variables There are two measures of obesity available, based on the two most widely used reference
panels – the International Obesity Task Force (IOTF) (Cole et al., 2000) and the UK90 (Cole
et al., 1990). Cut-offs are based on the age of the cohort member at the time of interview and
are shown below.
Variable Description
FCOVWGT6 Overweight cut-off point for CM's age and sex (IOTF thresholds)
FCOBESE6 Obesity cut-off point for CM's age and sex (IOTF thresholds)
FCUNDWU6 Underweight cut-off point for CM's age and sex (UK90 2nd centile)
FCOVWTU6 Overweight cut-off point for CM's age and sex (UK90 85th centile)
FCOBESU6 Obesity cut-off point for CM's age and sex (UK90 95th centile)
FCMCS6AG Age at interview to nearest 10th of year
FCBMIN6 MCS6 body mass index calculated (CLS)
FCOBFLG6 MCS6 obesity flag - IOTF thresholds
FCUK90O6 MCS6 obesity flag - UK90 thresholds
10.9 Reference cut-offs The following cut-offs were used for the construction of the IOTF and UK90 derived variables.
For the UK90 derivation cut-off points were generated using the LMSGrowth Microsoft Excel
add-in software.
40
IOTF BMI cut-offs
Age IOTF BMI cut-offs
Boys Girls
Overweight Obese Overweight Obese
13 21.910 26.840 22.580 27.760
13.1 21.982 26.922 22.660 27.848
13.2 22.054 27.004 22.740 27.936
13.3 22.126 27.086 22.820 28.024
13.4 22.198 27.168 22.900 28.112
13.5 22.270 27.250 22.980 28.200
13.6 22.340 27.326 23.052 28.274
13.7 22.410 27.402 23.124 28.348
13.8 22.480 27.478 23.196 28.422
13.9 22.550 27.554 23.268 28.496
14 22.620 27.630 23.340 28.570
14.1 22.688 27.700 23.404 28.630
14.2 22.756 27.770 23.468 28.690
14.3 22.824 27.840 23.532 28.750
14.4 22.892 27.910 23.596 28.810
14.5 22.960 27.980 23.660 28.870
14.6 23.026 28.044 23.716 28.918
14.7 23.092 28.108 23.772 28.966
14.8 23.158 28.172 23.828 29.014
14.9 23.224 28.236 23.884 29.062
15 23.290 28.300 23.940 29.110
15.1 23.352 28.360 23.986 29.146
15.2 23.414 28.420 24.032 29.182
15.3 23.476 28.480 24.078 29.218
15.4 23.538 28.540 24.124 29.254
15.5 23.600 28.600 24.170 29.290
15.6 23.660 28.656 24.210 29.318
15.7 23.720 28.712 24.250 29.346
15.8 23.780 28.768 24.290 29.374
15.9 23.840 28.824 24.330 29.402
16 23.900 28.880 24.370 29.430
41
UK90 BMI cut-offs
Age UK90 BMI cut-offs
Boys Girls
Underweight Overweight Obese Underweight Overweight Obese
13 14.774 20.652 22.768 14.939 21.742 24.056
13.1 14.822 20.726 22.850 14.99 21.815 24.135
13.2 14.87 20.799 22.931 15.041 21.889 24.215
13.3 14.918 20.873 23.012 15.091 21.961 24.293
13.4 14.967 20.947 23.012 15.141 22.033 24.371
13.5 15.017 21.021 23.174 15.192 22.104 24.448
13.6 15.066 21.095 23.255 15.241 22.175 24.524
13.7 15.115 21.169 23.335 15.29 22.244 24.599
13.8 15.166 21.243 23.416 15.339 22.313 24.673
13.9 15.215 21.317 23.416 15.387 22.381 24.745
14 15.266 21.391 23.577 15.436 22.449 24.819
14.1 15.317 21.466 23.657 15.484 22.516 24.89
14.2 15.367 21.539 23.737 15.53 22.581 24.96
14.3 15.417 21.613 23.816 15.577 22.646 25.028
14.4 15.468 21.686 23.895 15.623 22.709 25.097
14.5 15.518 21.759 23.973 15.669 22.773 25.164
14.6 15.569 21.833 24.052 15.714 22.835 25.23
14.7 15.619 21.905 24.128 15.758 22.896 25.295
14.8 15.67 21.978 24.207 15.802 22.956 25.359
14.9 15.72 22.049 24.283 15.846 23.016 25.423
15 15.77 22.121 24.359 15.889 23.075 25.485
15.1 15.82 22.193 24.436 15.931 23.133 25.546
15.2 15.87 22.264 24.511 15.973 23.189 25.605
15.3 15.921 22.335 24.587 16.013 23.244 25.664
15.4 15.969 22.404 24.660 16.054 23.3 25.722
15.5 16.02 22.475 24.735 16.094 23.353 25.778
15.6 16.069 22.544 24.808 16.133 23.406 25.835
15.7 16.118 22.613 24.880 16.171 23.457 25.888
15.8 16.167 22.682 24.953 16.21 23.509 25.943
15.9 16.215 22.749 25.023 16.247 23.56 25.996
16 16.264 22.817 25.094 16.284 23.609 26.048
42
11. Accelerometry
11.1 Background and introduction Accelerometry is the objective measurement of physical activity using accelerometers, which
are electro-mechanical devices that measure acceleration force. Cohort members were asked
to wear wrist-worn accelerometers for two days as part of the Age 14 Survey.
This is the second time accelerometry has been included on MCS; cohort members were
asked to wear a hip-worn accelerometer as part of the Age 7 Survey.
At age 14, all cohort members in Scotland, Wales and Northern Ireland were asked to wear
one, and an 81 per cent random subsample of respondents in England (due to resource
constraints). It was deemed important to have sufficient sample sizes to allow for country-
specific analysis. Further information about the implementation of accelerometry in the MCS
Age 14 Survey can be found in the MCS6 Technical Report.
11.2 Device use The device used at the Age 14 Survey was the GENEActiv Original (‘GENEActiv’), a wrist-
worn triaxial accelerometer. Measurement was obtained at 40Hz, chosen after consulting the
relevant literature and experts. This means there are 120 data points for every second the
device is recording (40 measurements per second on each of three directional axes). The
GENEActiv had a battery capable of lasting for the duration of the period in which
measurement would be required, and also had sufficient internal memory to record data for
that length of time. The device, which was robust and waterproof, did not provide respondents
with any feedback during the time they were wearing it.
Further technical information can be found on the GENEActiv website:
http://www.geneactiv.org/.
11.3 Subsampling All cohort members in Wales, Scotland and Northern Ireland were included in the subsample,
along with approximately 81 per cent of cohort members in England.
Numbers and proportions of cohort members in each country included in the subsample
Number Percent
England 6,290 80.3
Scotland 1,277 100
Wales 1,636 100
Northern Ireland 1,134 100
TOTAL 10,337 86.9
11.4 Days worn Cohort members were asked to wear the activity monitor for two randomly selected 24-hour
periods in the ten days following the visit (with the day of the visit and the following two days
ineligible for selection). One selected day was a weekday; the other, a weekend day. The days
were selected by the CAPI programme during the interviewer visit. The two selected days
were the same as those for the time-use diaries.
11.5 Consent As cohort members were still minors, parents were gatekeepers to their participation.
Information about the activity monitoring task was provided to both the parents and young
43
people in advance of the interviewer visit in the form of a leaflet, which contained information
on why the survey included an activity-monitoring component and briefly explained that
physical activity data would be collected using a wrist-worn monitor. Parents were asked to
provide written consent for the interviewer to approach their child to participate, and young
people gave verbal consent using a structured consent form, which the interviewer read out,
and then signed.
11.6 Baseline numbers Initial numbers on response are available in the MCS6 Technical Report. Further information
on valid data extracted from the devices, wear time and activity levels will be available when
the data is processed and released to the UK Data Service (later 2017).
44
12. Time-use diary
12.1 Background and introduction The time-use diary aimed to collect information about the daily activities of cohort members,
as well as contextual information including where they were, who they were with, and how
much they liked each activity. Cohort members were asked to complete the time-use diary for
two days as part of the Age 14 Survey. This is the first time MCS cohort members have been
asked to complete a time-use diary.
The time-use diary task was paired with the accelerometry element, with cohort members
wearing the monitor and completing the time-use diary on the same two days. Due to resource
constraints, a random subsample of respondents was invited to wear the accelerometer and
complete the time-use diary. Information about the subsample can be found in the
accelerometry section of this report (section 11).
The development of the MCS time diary instruments was led by the Centre for Longitudinal
Studies (CLS) in collaboration with Ipsos MORI and the Centre for Time Use Research
(CTUR) at the University of Oxford. CLS oversaw and contributed to all aspects of the
development. IM produced the time diary instruments and leaflets and carried out the different
testing phases. CTUR contributed significantly to the instrument development, advising on key
research design and implementation decisions.
Further information about the development of the time-use diary for the MCS Age 14 Survey
can be found in CLS working paper 2015/5, ‘Measuring young people’s time-use in the UK
Millennium Cohort Study: A mixed-mode time diary approach’, available at
http://www.cls.ioe.ac.uk/page.aspx?&sitesectionid=939&sitesectiontitle=CLS+working+paper
+series.
12.2 Days completed Cohort members were asked to complete the time-use diary on the same two randomly
selected days on which they wore the activity monitor. These were two randomly selected 24-
hour periods in the ten days following the visit (with the day of the visit and the following two
days ineligible for selection). One selected day was a weekday; the other, a weekend day.
The days were selected by the CAPI programme during the interviewer visit.
12.3 Modes The time-use diary was available in three modes – web, app and paper. Young people were
encouraged to use the web or app diary and were only offered the paper version if they could
not, or declined to, complete the diary using one of the electronic methods. The web and paper
versions used the conventional light diary time grid, which asked young people to report what
they were doing every 10 minutes across the selected 24 hour period. The app used a
question-based approach for reporting activities and contextual information. The following
table sets out the similarities and differences between the diaries when administered in the
three different modes.
Similarities and differences between the time diary modes
Web App Paper
Approach Time-grid Question-based Time-grid
Time unit 10-minute slot User-assigned start and end times
10-minute slot
Diary dimensions Overlap Coterminous Overlap
45
Soft and hard checks
Yes Yes No
Aide-memoire Yes Yes No
12.4 Consent As the cohort members were minors, parents were gatekeepers to their participation.
Information about the time-use task was provided to both the parents and young people in
advance of the interviewer visit in the form of a leaflet, which contained information on why
the survey included the collection of time-use information and briefly explained how the
information would be collected. Parents were asked to provide written consent for the
interviewer to approach their child to participate, and young people gave verbal consent using
a structured consent form, which the interviewer read out, and then signed.
12.5 Baseline numbers Initial numbers on response are available in the Technical Report. Further information will be
available when the data is processed and released to the UK Data Service
46
13. Saliva
13.1 Background and introduction Cohort members and resident biological parents were asked to provide a saliva sample as
part of the Age 14 Survey. The purpose of this collection was to extract DNA for later
genotyping for future research.
Saliva collection has been included in previous sweeps of MCS, though on a smaller scale
and for a different purpose. At the Age 3 Survey, a saliva sample was taken from the children
to measure exposure to common childhood infections. The saliva was not used for DNA or
genetic testing. At the Age 11 Survey, collection of saliva for DNA extraction and storage was
piloted, but not carried out at the main stage due to lack of funding.
13.2 Saliva collection protocols All cohort members who lived with a parent or guardian who had legal parental responsibility
were eligible to provide a saliva sample. This is because a person with legal parental
responsibility was required to provide written consent for the cohort member to give a sample.
In addition, biological parents who lived with the cohort member were eligible to provide a
saliva sample.
Saliva was collected using Oragene 500 DNA self-collection kits. Respondents were required
to spit into a container until they had provided 2ml of saliva. The interviewer then closed the
lid of the container, releasing preservative into the saliva sample. Samples were posted by
interviewers to the laboratory appointed to extract and store DNA, at the University of Bristol.
More information about the collection kit used can be found on the DNA Genotek website:
http://www.dnagenotek.com/ROW/products/OG500.html
13.3 Consent Information about the saliva collection task was provided to both the parents and young people
in advance of the interviewer visit in the form of a booklet, which contained information about
why the study wanted to collect saliva and what it would be used for. Parents were asked to
provide written consent to provide their own sample, and parents with legal parental
responsibility provided consent for the cohort member to give a sample. Young people gave
verbal consent to provide a sample using a structured consent form, which the interviewer
read out, and then signed.
13.4 Storage The saliva samples were sent to a research laboratory at the University of Bristol, where DNA
was extracted and is being stored in freezers. Researchers wishing to use the DNA samples
will be required to apply to the METADAC. More information about the METADAC and the
application process will be made available on the website when data access arrangements
have been established: http://www.metadac.ac.uk/.
47
13.5 Baseline numbers The table below gives indicative number of samples expected to be available. The actual
number of samples will be lower due to losses in DNA extraction. Final figures will be made
available once the samples have been fully processed.
Number of saliva samples obtained
Number
Main respondent 9,584
Partner respondent 5,169
Cohort member 9,726
48
14. Response and weights at Sweep 6 As with any longitudinal survey, the MCS is subject to attrition. Attrition takes place when
respondents drop out of the survey over time. This leads to two problems:
(i) a reduction in sample size, and
(ii) bias in sample composition.
Sample bias arises when the likelihood of dropping out from the survey is correlated with the
socio-demographic characteristics of the respondents. In this case, the survey will lose a
particular type of respondent (e.g., disadvantaged families and ethnic minorities) and the
sample will no longer be representative of the population it was drawn from. However, there
are statistical methods to deal with this so as to ensure the remaining sample recovers (under
reasonable assumptions) population parameters, which are the topic of this section.
This section examines attrition in sweep 6 of MCS and presents the procedures used in the
construction of MCS6 unit non-response weights. For a full description of attrition in previous
sweeps, refer to the MCS Technical Report on Response (3rd edition, 2010) and Technical
Report on Response in Sweep 5 (2014). For a description of how to use the weights in Stata
and SPSS, refer to the respective guides (Stata, SPSS). For a description of the MCS sample
refer to the Technical Report on Sampling (4th edition, 2007). All these documents are on the
MCS Survey Design pages of the CLS website7.
14.1 Response in MCS In Table 1, the proportions of productive and unproductive cases are presented by category.
The proportion of productive cases decreased over time from 96.4 per cent in MCS1 to 60.9
per cent in MCS6. The two categories of non-response which have seen a marked increase
over time are ‘refusal’ and ‘not issued’. ‘Refusals’ consist of respondents who declined to take
part in a particular sweep of data collection, and ‘not issued’ are respondents who have not
participated in the survey on two consecutive occasions, and therefore were no longer issued
for fieldwork (i.e., the survey agency no longer tries to contact them). Non-contact has declined
over time because respondents in this category have either been located and contacted again,
or have moved to the not issued category. All other types of non-response are relatively stable
over time. Note that ‘ineligible’ includes child deaths, sensitive cases and temporary and
permanent emigrants. The category ‘untraced movers’ refers to respondents who have
changed address and were not located, including possible emigrants. Respondents who were
not issued in MCS1 are labelled as ‘new families’. These were eligible families who were not
contacted in MCS1 because their addresses were not know in time for them to be included in
the first wave of data collection.
7 http://www.cls.ioe.ac.uk/page.aspx?&sitesectionid=880&sitesectiontitle=Survey+Design
49
Table 1: Productive and unproductive cases in all MCS sweeps
MCS1 MCS2 MCS3 MCS4 MCS5 MCS6
Age 9 months
Age 3 years
Age 5 years
Age 7 years
Age 11 years
Age 14 years
Freq. % Freq. % Freq. % Freq. % Freq. % Freq. %
Productive 18,551 96.4 15,590 81.0 15,246 79.2 13,857 72.0 13,287 69.0 11,726 60.9
Refusal 1,739 9.0 2,315 12.0 1,811 9.4 2,195 11.4 3,029 15.7
Ineligible 167 0.9 300 1.6 126 0.7 78 0.4 45 0.2
Untraced movers 686 3.6 546 2.8 706 3.7 388 2.0 428 2.2
Non-contact 930 4.8 546 2.8 123 0.6 438 2.3 75 0.4
Not issued 692 3.6 2,212 11.5 2,851 14.8 3,828 19.9 Other unproductive 131 0.7 290 1.5 408 2.1 6 0.0 112 0.6
Total 19,243 100 19,243 100 19,243 100 19,243 100 19,243 100 19,243 100
Figure 1 presents the proportion of productive cases in MCS in all sweeps, showing that the
sample decreased by 40 per cent by the time of the Age 14 Survey.
Figure 1: Proportion of productive cases in all MCS sweeps
Note: The total number of MCS respondents ever interviewed is 19,243.
We next show how the proportion of productive cases at MCS6 varies along key dimensions.
First, Table 2 shows how the MCS6 proportion of productive cases varies by country of
sampling. The productive proportion is higher than the UK average in England, while it is lower
than the average in Scotland and Northern Ireland.
96
81 7972 69
61
0102030405060708090
100
Age 9
months
Age 3
years
Age 5
years
Age 7
years
Age 11
years
Age 14
years
Per
cen
t
50
Table 2: Productive and unproductive cases by country of sampling in MCS6
England Wales Scotland Northern Ireland
Freq. % Freq. % Freq. % Freq. %
Productive 7,678 62.8 1,669 60.5 1,263 54.1 1,116 58.0
Refusal 1,853 15.2 455 16.5 401 17.2 320 16.6
Ineligible 35 0.3 2 0.1 6 0.3 2 0.1
Untraced 199 1.6 92 3.3 101 4.3 36 1.9
Non-contact 60 0.5 6 0.2 6 0.3 3 0.2
Not issued 2,335 19.1 503 18.2 548 23.5 442 23.0 Other unproductive 64 0.5 33 1.2 11 0.5 4 0.2
Total 12,224 100 2,760 100 2,336 100 1,923 100 Sample size=19,243. Percentages may not sum to 100 due to rounding.
Table 3 shows that the proportion of productive cases varies across sampling strata in each
country. Respondents sampled from the socially advantaged stratum are more likely to be
productive in all four countries, compared to those sampled from the disadvantaged stratum.
Respondents sampled from the ethnic minority stratum are less likely to be productive than
those in the advantaged stratum in England.
Table 3: Proportion of productive cases by stratum in MCS6
England Wales Scotland Northern Ireland
Adv. Dis. Ethn. Adv. Dis. Adv. Dis. Adv. Dis.
Productive 67.1 59.8 60.3 65.1 58.5 60.1 48.3 63.4 54.8
Unproductive 32.9 40.2 39.7 34.9 41.6 39.9 51.7 36.7 45.2 Note: Adv. stands for advantaged stratum. Dis. stands for disadvantaged stratum and Ethn. stands for ethnic
minority stratum. Sample size=19,243.
In Table 4 we look at different response patterns. Table 4 shows that 47.2 per cent of all
respondents participated in all six sweeps of MCS. In contrast, 22.1 per cent have interrupted
response patterns (i.e., non-monotone response). In other words, they participated in a
number of sweeps, and then dropped out before participating again in subsequent sweeps.
30.7 per cent of all respondents have monotone response patterns. That is, they participated
in a number of sweeps before dropping out for all subsequent sweeps.
Table 4: Monotone vs. non-monotone responses in all MCS sweeps
Patterns Freq. Percent
Monotone 5,908 30.7
Non-monotone 4,247 22.1
All waves 9,088 47.2
Total 19,243 100
Table 5 shows the percentages of respondents participating in n sweeps (n=1…6). We see
that 64.1 per cent of respondents participated in at least five out of six sweeps of MCS, which
indicates that more than half of the sample have almost complete records.
51
Table 5: Number of times productive up to MCS6
Times productive Freq. Percent
One 1,932 10.0
Two 1,391 7.2
Three 1,559 8.1
Four 2,027 10.5
Five 3,246 16.9
Six 9,088 47.2
Total 19,243 100
Note: Percentages may not sum to 100 due to rounding.
14.2 Predicting response at MCS6 The procedure used for predicting responses at Sweep 5 was also used at Sweep 6 8. We
estimate a logit model in which the dependent variable is binary (=1 for response and 0
otherwise). The predictors are:
1. The cohort member’s gender
2. Mother’s age at first live birth
3. The cohort member’s ethnic group
4. Housing tenure in MCS5
5. Accommodation type in MCS5
6. The main respondent’s highest educational qualification in all sweeps
7. Whether the cohort member was breastfed
8. Number of parents living in the household in MCS5
9. The main respondent’s highest social and economic status in all sweeps
10. Ratio of number of times not answering the income question divided by the number of
sweeps productive
11. Ratio of number of times reporting having a job divided by the number of times
productive
12. Whether the household is a ‘new’ family. There were 701 children who joined the
survey in Sweep 2 because their addresses were not known in Sweep 1; therefore
they did not take part in the first sweep. These children and their families were labelled
as ‘new families’.
Missing data for predictor variables due to non-monotone non-response or item missingness
were imputed using simple and multiple imputations, as described in Appendix A.
Table A1 in the Appendix shows the odds ratios of the response logit model estimated using
the 25 imputed datasets. The linear predicted values were generated from this model; then
the predicted values were converted into predicted probabilities using an inverse logit
transformation. The non-response weights for Sweep 6 were constructed as the inverse of the
predicted probabilities [Wooldridge 2007]. Two overall weights were constructed by multiplying
the sampling weights in Sweep 1 by the attrition weights in each sweep of MCS (i.e., 2 to 6).
The weights were scaled to make their total equal to the productive sample size. We note that
the effectiveness of the response weights to correct for bias depends on the inclusion of all
8 http://eprints.ioe.ac.uk/18759/1/Technical_Report_on_Response_in_Sweep5_for_web_TM.pdf
52
important predictors of unit non-response in the logit response model (Table A1) (Seaman and
White 2013).
Weights variable names:
FOVWT1: Sweep 6 overall weight for single country analysis
FOVWT2: Sweep 6 overall weight for whole of UK analysis.
In Tables A2 and A3 of the Appendix, the means, minima and maxima of the two weights are
presented by stratum.
For a description of how to use the weights in Stata and SPSS, refer to the respective guide:
Stata, SPSS. Available on the MCS Survey Design pages of the CLS website 9.
9 http://www.cls.ioe.ac.uk/page.aspx?&sitesectionid=880&sitesectiontitle=Survey+Design
53
15. Available datasets
There are two sets of 13 data files (one set of SPSS data files and one of STATA data files):
1) A longitudinal file containing information on the family which is consistent over time or is
the most current version of a longitudinal variable:
MCS Longitudinal Family Level Information.
This file contains one row for all families in the longitudinal sample: i.e., families who
have taken part in MCS1 or MCS2 (n=19,243 (18,552+691)).
2) The cross-sectional data from the household questionnaire, main, partner and proxy
interviews:
MCS6 parent interview data
MCS6 parent interview unfolding brackets data (in preparation)
MCS6 proxy parent interview data
MCS6 proxy parent interview unfolding brackets data (in preparation)
This file contains one row per respondent.
3) Cross-sectional household grid data:
MCS6 Household Grid
These files contain one row for each person (ever) in the household grid in productive
families.
4) Child questionnaire files
MCS6 child interview data
MCS6 parent CM interview data
5) Child assessment and measurement files:
MCS6 child assessment data
MCS6 child measurement data
MCS6 accelerometer data (in preparation)
MCS6 time-use diary (in preparation)
6) Parental assessment (word activity)
MCS6 parent assessment
7) Geographically linked data (IMD and rural/urban indicators):
MCS6 IMDRU England
MCS6 IMDRU Wales
MCS6 IMDRU Scotland
MCS6 IMDRU Northern Ireland
8) Derived variables:
MCS6 family level derived variables.
These files contain one row for each productive family at that sweep.
54
MCS6 parent derived variables
MCS6 CM derived variables
These files contain one row for each productive main, partner or CM respondent at that sweep.
Data can be downloaded from the UK Data Service at
https://discover.ukdataservice.ac.uk/series/?sn=2000031
55
16. Accompanying documents
16.1 Questionnaires
MCS6 CAPI questionnaire
MCS6 Young person questionnaire
MCS6 Physical measurement activity monitor time use
MCS6 Age 14 survey SDQ
16.2 Technical Reports
MCS6 Technical Report
MCS6 Technical Report Appendix A Pre-notification and briefing docs
MCS6 Technical Report Appendix B Household materials
MCS6 Technical Report Appendix C Welsh and ethnic minority languages
MCS6 Technical Report Appendix D
16.3 Guides
MCS6 Time-Use diary documentation
MCS6 Technical Report on response
User Guide to analysing MCS data using Stata
User Guide to analysing MCS data using SPSS
MCS6 Derived Variables User Guide
56
17. Appendix A: Response and weights
Imputations Multiple imputations were carried out using the MI command in Stata 13. As a result of the
use of simple and multiple imputations, the sample used in the logit response model consisted
of 15,415 observations (i.e., the issued sample in MCS6). Weights were constructed for all
respondents in MCS6. Imputations were carried out as described in Appendix I.
Replacement of missing values Since ethnicity is a fixed attribute over time and the main respondent’s highest educational
qualification is unlikely to change from one sweep to the other, we replaced the missing values
on these two variables in MCS6 using the most recent available information from previous
sweeps. Mother’s age at first live birth was missing for only 60 cases. These were replaced
by the average of non-missing cases.
Multiple imputations Four variables were imputed using multiple imputation with chained equations in Stata 13. The
imputation was carried out for item missingness and missingness caused by non-monotone
response patterns. Accommodation type had the largest proportion missing with 22 per cent.
We generated 25 multiple imputations. The number of parents in the household and housing
tenure were missing for 2,177 respondents in the analytical sample (i.e., issued sample in
MCS6). Whether the cohort member was breastfed was missing for 612 respondents – mostly
new families who joined the survey in Sweep 2 – and accommodation type was missing for
3,435 respondents.
Breastfeeding and type of accommodation were imputed using the following variables as
predictors of item specific missingness: highest educational qualification, whether family is a
new family, cohort member’s gender, cohort member’s ethnic group, the main respondent’s
highest social and economic status, and sampling stratum.
The number of parents living in the household was imputed using highest educational
qualification, whether family is a new family, cohort member’s gender, cohort member’s ethnic
group, the main respondent’s highest social and economic status, sampling stratum, in
addition to the number of parents in the household in MCS1 and MCS4.
Housing tenure was imputed using highest educational qualification, whether family is a new
family, cohort member’s gender, cohort member’s ethnic group, the main respondent’s highest
social and economic status, and sampling stratum, in addition to housing tenure in all previous
sweeps.
We note that multiple imputation returns valid estimates assuming the data are Missing at
Random (MAR) (Enders (2010); Seaman et al., (2013); Sterne et al., (2009). This implies that
any differences between the missing values and the observed values can be explained by the
variables that were included in the imputation models. Put differently, conditional on the
variables in the imputation model, missingness is not due to unobserved or observed variables
not included in the model.
57
No imputation required The ratio of number of times the income question was not answered divided by the number of
productive sweeps did not have any missing data since it was constructed using non-missing
observations from all sweeps. The same applies for the ratio of number of times the informant
reported having a job divided by the number of times productive. The main respondent’s
highest social and economic status was constructed as the maximum of social and economic
status reported in each sweep.
Finally, some variables, such as cohort member’s gender and whether the household is a new
family, did not have any missing values and therefore did not require any imputation.
Table A1: Logit response model
Explanatory variables Odds ratio SE T P>t [95% Conf.
Interval]
New family 1.03 0.11 0.28 0.78 0.84 1.26
Cohort member is a boy 0.87 0.03 -3.49 0.00 0.81 0.94
Cohort member's ethnic group (Reference: White)
Mixed 0.85 0.10 -1.42 0.15 0.68 1.06
Indian 1.35 0.19 2.17 0.03 1.03 1.78
Pakistani, Bangladeshi 2.01 0.19 7.31 0.00 1.67 2.43
Black 0.79 0.09 -2.19 0.03 0.64 0.97
Other, NA, not known, refusal 1.22 0.17 1.43 0.15 0.93 1.61
Main respondent's highest social and economic status (Reference: Managerial and Professional)
Intermediate 0.83 0.05 -2.87 0.00 0.74 0.94
Small employers and self-employed 0.77 0.07 -2.65 0.01 0.64 0.94
Lower supervisory and technical 0.66 0.06 -4.47 0.00 0.54 0.79
Semi-routine and routine 0.52 0.03 -9.81 0.00 0.46 0.59
NA 0.45 0.05 -7.96 0.00 0.37 0.55
Highest educational qualification (Reference: NVQ level 1)
NVQ level 2 1.00 0.08 -0.05 0.96 0.85 1.16
NVQ level 3 1.11 0.10 1.21 0.23 0.94 1.32
NVQ level 4 1.35 0.12 3.46 0.00 1.14 1.60
NVQ level 5 1.84 0.22 5.16 0.00 1.46 2.33
Overseas qualifications only 1.22 0.16 1.47 0.14 0.94 1.58
None of these 0.80 0.07 -2.60 0.01 0.67 0.95
Number of parents in household (reference: One parent)
Two parents/carers 1.20 0.06 3.70 0.00 1.09 1.32
Child was breastfed at least once 1.23 0.06 4.49 0.00 1.12 1.35
Accommodation type (Reference: Other)
House or bungalow 1.14 0.11 1.39 0.17 0.95 1.37
Housing tenure (Reference: Own outright)
Own – mortgage/loan 1.09 0.10 0.94 0.35 0.91 1.32
Rent from local authority 0.73 0.08 -2.96 0.00 0.59 0.90
Rent from housing association 0.72 0.08 -2.87 0.00 0.57 0.90
Rent privately 0.74 0.08 -2.79 0.01 0.60 0.91
Other (rent free, living with parents) 0.67 0.09 -3.02 0.00 0.51 0.87
58
Ratio income item non-response 0.26 0.03 -
11.43 0.00 0.20 0.32
Ratio times having a job 0.33 0.03 -
14.08 0.00 0.29 0.39
Mother's age at first birth 1.05 0.00 10.68 0.00 1.04 1.06
Constant 1.81 0.36 2.96 0.00 1.22 2.69
N 15,415 Note: The analytical sample in table 6 includes all issued cases in MCS6 with 15,415 observations.
Table A2: FOVWT1, Sweep 6 overall weight for single country analysis
Sampling stratum N Mean Std. Dev. Min Max
England – advantaged 3,240 1.24 0.66 0.68 8.96
England – disadvantaged 2,876 1.00 0.68 0.38 10.90
England – ethnic 1,562 0.42 0.30 0.14 4.28
Wales – advantaged 542 1.54 0.87 0.79 10.80
Wales – disadvantaged 1,127 0.79 0.49 0.32 6.93
Scotland – advantaged 688 1.06 0.79 0.32 7.95
Scotland – disadvantaged 575 0.95 0.81 0.20 9.23
Northern Ireland – advantaged 458 1.25 0.86 0.34 6.07
Northern Ireland – disadvantaged 658 1.03 0.88 0.20 8.45
All strata 11,726 1.01 0.73 0.14 10.90
Table A3: FOVWT2, Sweep 6 overall weight for whole of the UK analysis
Sampling stratum N Mean Std. Dev. Min Max
England – advantaged 3,240 1.60 0.88 0.86 12.47
England – disadvantaged 2,876 1.31 0.91 0.49 14.03
England – ethnic 1,562 0.56 0.40 0.18 5.53
Wales – advantaged 542 0.52 0.29 0.27 3.48
Wales – disadvantaged 1,127 0.27 0.16 0.11 2.37
Scotland – advantaged 688 0.82 0.59 0.25 6.26
Scotland – disadvantaged 575 0.74 0.63 0.16 7.30
Northern Ireland – advantaged 458 0.49 0.32 0.14 2.43
Northern Ireland – disadvantaged 658 0.40 0.33 0.08 3.33
All strata 11,726 1.01 0.86 0.08 14.03
18. References Calderwood, L., Smith, K., Gilbert, E., Rainsberry, M., Knibbs, S., and Burston, K. (Winter 2015). Securing participation and getting accurate answers from teenage children in surveys: Lessons from the UK Millennium Cohort Study. Social Research Practice, 1. Cole T.J., Bellizzi M.C., Flegal K.M., Dietz W.H. (2000). Establishing a standard definition for child overweight and obesity worldwide: International survey. BMJ, 320:1240. Cole T.J., Freeman J.V., Preece M.A. (1995). Body mass index reference curves for the UK, 1990. Archives of disease in childhood, 73(1): 25-9.
59
Conroy, R.M. (2005). Stings in the tails: Detecting and dealing with censored data. Stata Journal, 5, 395-404. DWP (2016). Households below average income. 1994/95-2014/15. Retrieved from https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/532416/households-below-average-income-1994-1995-2014-2015.pdf
Enders, C. K. (2010). Applied missing data analysis, Guilford Press.
Gilbert, E., Calderwood, L., Rainsberry, M., Knibbs, S., and Burston, K., (2015). ‘Tweets, branding and swag: Engaging teenagers in research’. European Survey Research Association Conference, Reykjavik. Goodman, R (1997). The Strengths and Difficulties Questionnaire: A research note. Journal of Child Psychology and Psychiatry, 38, 581-6; Psychological and Development Inventories. Hansen, K. (2008). Millennium Cohort Study First, Second and Third Surveys: A Guide to the
Datasets, Third Edition. Retrieved from www.cls.ioe.ac.uk/shared/get-
file.ashx?id=598&itemtype=document
HBAI Team, Information Directorate, Department for Work and Pensions (2005). Households
below average income statistics: Adoption of new equivalence scales.
(http://www.dwp.gov.uk/asd/hbai/nsfr_newequiv.pdf;). Accessed 22 May 2008.
Johnson, J., Atkinson, M., Rosenberg, R. (2015). ‘Millennium Cohort Study: Psychological, developmental and health inventories’. (Jan 2015), CLS User Guide to the Data.
Kessler, R.C., Barker, P.R., Hirpi, E., Howes, M. J, Normand, S-L. T., Manderscheid, R.W.,
Walters, E.E., Zaslavsky, A.M (2003) Screening for serious mental illness in the general
population, Archives of General Psychiatry 60(2) 184-9.
Pan, H. and Cole, T.J. (2012). LMSgrowth, a Microsoft Excel add-in to access growth references based on the LMS method. Version 2.77. http://www.healthforallchildren.co.uk/
Plewis, I. (Ed). (2007). Millennium Cohort Study First Survey: Technical report on sampling
(4th edition). London: Centre for Longitudinal Studies, Institute of Education.
Seaman, S., Galati, J., Jackson, D. and Carlin, J. (2013). What is meant by ‘missing at
random’? Statistical Science, 28, 257-68.
Seaman, S. R. and White, I. R. (2013). Review of inverse probability weighting for dealing with
missing data. Stat Methods Med Res, 22, 278-95.
Smith, K., Calderwood, L., Knibbs, S., and Burston, K. (2015). ‘Let’s talk about sex: asking 14-
year-olds about their lives in a home setting.’ European Survey Research Association
Conference, Reykjavik.
StataCorp. (2007). Stata Statistical Software: Release 10. College Station, TX: StataCorp Lp.
Sterne, J. A. C., White, I. R., Carlin, J. B., Spratt, M., Royston, P., Kenward, M. G., Wood, A.
M. and Carpenter, J. R. (2009). Multiple imputation for missing data in epidemiological and
clinical research: potential and pitfalls. BMJ, 338.
60
Stewart, M.B. (1983). On least-squares estimation when the dependent variable is grouped.
Review of Economic Studies, 50, 737-53.
Wooldridge, J. M. (2007). Inverse probability weighted estimation for general missing data
problems. Journal of Econometrics, 141, 1281-1301.